measurement and modelling of water use of citrus …
TRANSCRIPT
MEASUREMENT AND MODELLING OF WATER USE OF CITRUS ORCHARDS
By
WALTER MAHOHOMA
A thesis submitted in fulfilment of the requirements for the degree of PhD in
Agronomy
In the Faculty of Natural and Agricultural Sciences
University of Pretoria
Supervisor: Dr NJ Taylor
Co-Supervisor: Prof. JG Annandale
April 2016
© University of Pretoria
i
Acknowledgements
My sincere gratitude goes to my supervisors Dr NJ Taylor and Professor JG
Annandale for the guidance throughout the course of my study. The CSIR staff who
worked on the “Water use of fruit trees” project who are Dr MB Gush, Dr AD Clulow,
Mr V Naiken and Dr MG Mengistu for all the assistance. I am also thankful to my
colleagues in the water group in the department for all the assistance. I am also
grateful to Dr S Dzikiti for his advice throughout the writing of this thesis.
The project would not be possible without the understanding of Moosrivier Farm
(Schoeman Boerdery Group) staff who provided the facilities for carrying out this
project especially Messrs J Burger, H Schoeman and B du Toit we worked with.
My sincere gratitude also goes to all the staff at the University of Pretoria Farm for
the support especially Mr L Nonyane and Mr L van Merwe. Sencere gratitude to my
colleagues Nadia Ibramo, Solomon Ghezehei, Melake Fessehazion and Godwill
Madamombe for the support. I also would like to thank members of the Agronomy
Department at Midlands State University for the support.
I gratefully acknowledge the Water Research Commission (WRC) and South African
Department of Agriculture for providing all the financial support throughout the
course of my study.
Last but not least much love goes to my wife (Tsungai), two brothers (Webster and
Wellington), and two sisters (Sekai and Longina (RIP)), together with their families
for their unwavering support.
© University of Pretoria
ii
Declaration
I, Walter Mahohoma, declare that the dissertation, which I hereby submit for the
degree PhD Agronomy at the University of Pretoria, is my own work and has not
previously been submitted by me for a degree at another university. Where
secondary material is used, this has been carefully acknowledged and referenced in
accordance with university requirements. I am aware of university policy and
implications regarding plagiarism.
Walter Mahohoma
Date: 15 September 2016
© University of Pretoria
iii
Abstract
MEASUREMENT AND MODELLING OF WATER USE OF CITRUS ORCHARDS
By
Walter Mahohoma
Supervisor: Doctor NJ Taylor
Co-supervisor: Professor JG Annandale
Degree: PhD Agronomy
Water requirements of evergreen citrus orchards in semi-arid regions of the world
are mostly met through irrigation. Definitive knowledge of crop water use is the
fundamental basis for sound water management under these production systems.
The objectives of this study were two-fold: i) to measure long-term transpiration of
well managed micro-irrigated citrus orchards and; ii) to test physically–based models
that can be used to predict transpiration across different citrus growing regions.
Whole tree transpiration measurements were conducted using the heat ratio method
(HRM) sap flow technique in two well-irrigated citrus orchards [Citrus sinensis (L.)
Osbeck] in the summer rainfall area of South Africa. The two orchards planted with
‘Delta’ Valencia and ‘Bahianinha’ Navel orange trees were located at Moosrivier
Farm in Groblersdal. A data set from an orchard planted with ‘Rustenburg’ Navel
orange trees that was located in the winter rainfall area was sourced for model
validation. The orchard was located at Patrysberg Farm in the winter rainfall area. All
the orchards were drip irrigated and managed according to the industry standards.
Weather parameters were monitored using automatic weather stations that were
situated close to the fields, for modelling purposes.
The HRM was initially calibrated against eddy covariance measurements in winter
(31July to 3 August 2008) and autumn (21 to 25 May 2009) in the ‘Delta’ Valencia
orchard, when soil evaporation was considered negligible. Calibration was done by
adjusting the wounding width to ensure that average transpiration of sample trees
matched crop evapotranspiration (ETc) measured above the orchard using the eddy
covariance method. The wounding width, termed the ‘virtual wound width’, was
obtained by inputting ETc into regression equations of daily transpiration against
© University of Pretoria
iv
wound width. Xylem anatomical assessments were conducted on excised stem
samples prior to probe insertion and at the conclusion of sap flow measurements in
in order to ascertain and explain the virtual wound width determined through
calibration. The average virtual wound width that matched average transpiration of
sample trees to ETc in both winter and autumn was 3.2 mm. The similarity of the
virtual wound width obtained during winter and autumn suggests that wounding did
not increase with time and a single field calibration of the HRM using eddy
covariance measurements is sufficient for measuring long term transpiration in citrus
orchards.
The HRM was used to measure transpiration for periods of 364 days in the ‘Delta’
Valencia orchard, 301 days in the ‘Bahianinha’ Navel orchard and 365 days in the
‘Rustenburg’ Navel orchard. Average transpiration in the ‘Delta’ Valencia orchard
was 1.15 mm day-1 and 2.17 mm day-1 in winter and summer, respectively. On
average, ‘Bahianinha’ Navel orange trees transpired 0.77 mm day-1 and 1.69 mm
day-1 in winter and summer, respectively. In the ‘Rustenburg’ Navel orchard average
transpiration was 2.26 mm day-1 for summer and 2.08 mm day-1 for winter. Total
transpiration measured in the ‘Delta’ Valencia, ‘Bahianinha’ and ‘Rustenburg’ Navel
orchards was 650, 433 and 682 mm, respectively. Transpiration coefficients (Kt)
determined in the three orchards ranged between 0.28 and 0.71. The Kt values were
almost constant throughout the seasons in the summer rainfall area and significantly
higher in the winter months than in the summer months in the winter rainfall area.
Derived monthly and seasonal Kt determined for the three orchards were much lower
than the standardised values published in the FAO56. Differences among the
transpiration coefficients determined in the orchards in this study means that they
are orchard specific and can therefore not be directly applied or transferred to
different growing regions of the world.
There is a need to develop an easy method for estimating site-specific crop
coefficients, in order to improve water management in citrus orchards. The
determination of basal crop coefficients based on physical characteristics of the
vegetation and an adjustment for relative crop stomatal control over transpiration
formulated by Allen and Pereira (2009) was tested. Use of the parameters for
generic citrus trees suggested by the authors did not provide good estimates of
© University of Pretoria
v
transpiration in the three study orchards. A good agreement between measured and
estimated Kt values was obtained by back-calculating leaf resistances using
measured transpiration values, as recommended by Allen and Pereira (2009). The
values of leaf resistances obtained through this procedure were higher than the
values suggested by Allen and Pereira (2009), but comparable to those reported in
literature and measured in the ‘Rustenburg’ navel orchard. A relationship between
mean monthly leaf resistance and ETo in the ‘Rustenburg’ Navel orchard provided a
means of estimating mean leaf resistance which estimated Kt values with a
reasonable degree of accuracy in the three orchards. However, this relationship only
provided good seasonal estimates of transpiration, which means that it can only be
useful for irrigation planning.
A more mechanistic model capable of capturing the dynamics of the canopy
conductance in citrus varieties is required to predict transpiration of citrus trees on
day-to-day basis for purposes such as irrigation scheduling. The use of the Penman-
Monteith equation coupled with a Jarvis-type canopy conductance to predict long-
term transpiration was evaluated in the three orchards. Bulk canopy conductance
was calculated from the measured transpiration using a top-down approach by
inverting the Penman-Monteith equation. Response functions for solar radiation,
temperature and vapour pressure deficit were parameterised for a Jarvis model
based on the measured data. The Jarvis model was first parameterized in the ‘Delta’
Valencia orchard. On validation the approach predicted transpiration in the same
orchard reasonably well. However, transpiration was overestimated in the
‘Bahianinha’ orchard and underestimated in the ‘Rustenburg’ Navel orchard when
the Jarvis parameters obtained in the ‘Delta’ Valencia orchard were applied to the
two orchards. When the Jarvis model was parameterized in the ‘Bahianinha’ and
‘Rustenburg’ Navel transpiration was predicted fairly well. A single set of response
functions for canopy conductance obtained in each orchard was found to be able to
accurately predict transpiration in the respective orchard for a fairly long period. This
approach may be applicable to sub-tropical conditions where wetting of the canopy
by rainfall and occurrence of cloudy/overcast conditions are both infrequent.
© University of Pretoria
vi
Contents
Acknowledgements ..................................................................................................... i
Declaration .................................................................................................................. ii
Abstract ...................................................................................................................... iii
List of Figures ............................................................................................................. x
List of Tables ............................................................................................................. xv
List of abbreviations and symbols ........................................................................... xvii
Introduction and structure of thesis ............................................................................ 1
Aim of the study ...................................................................................................... 7
Thesis objectives .................................................................................................... 7
Hypotheses ............................................................................................................. 7
Thesis outline ......................................................................................................... 7
Chapter 1: Literature Review ...................................................................................... 9
1.1 Measurement of citrus water use ......................................................................... 9
1.1.1 Introduction .................................................................................................... 9
1.1.2 Sap flow measurements of crop water use .................................................. 13
1.1.2.1 Theory of heat pulse sap flow technique ............................................... 14
1.1.2.2 Compensation heat pulse method ......................................................... 15
1.1.2.3 Heat ratio method .................................................................................. 16
1.1.2.4 Correction for wounding ........................................................................ 17
1.1.2.5 Calibration of the heat pulse techniques ............................................... 18
1.1.2.6 Scaling from sample trees to orchard level ........................................... 19
1.1.3 Micrometeorological estimates of evapotranspiration .................................. 20
1.1.3.1 Eddy covariance method ....................................................................... 20
1.1.3.1.1 Direct measurement ........................................................................... 21
1.1.3.1.2 Energy balance method ..................................................................... 21
1.2 Modelling citrus water use .................................................................................. 23
1.2.1 Introduction .................................................................................................. 23
1.2.2 Crop coefficient approach ............................................................................ 24
1.2.2.2 Single crop coefficient ........................................................................... 25
1.2.2.3 Dual crop coefficient .............................................................................. 25
1.2.2.4 Reference evapotranspiration ............................................................... 25
1.2.2.5 Citrus crop coefficients .......................................................................... 27
© University of Pretoria
vii
1.2.3 Estimation of transpiration crop coefficients from fractional ground cover and
crop height ............................................................................................................ 31
1.2.3.1 Basal crop coefficient during peak plant growth (Kcb full) ........................ 32
1.2.3.2 Estimation of plant density coefficient (Kd) ............................................ 33
1.2.4 Modelling canopy conductance.................................................................... 36
1.2.4.1 Jarvis-type model .................................................................................. 37
1.2.4.3 Aerodynamic conductance .................................................................... 38
1.3 Scope of the study .............................................................................................. 39
Chapter 2: Calibration of the heat ratio method (HRM) sap flow technique for long-
term transpiration measurement in citrus orchards .................................................. 41
2.1 Introduction ........................................................................................................ 41
2.2 Materials and methods ....................................................................................... 43
2.2.1 Experimental Site and Climate ..................................................................... 43
2.2.2 Sap flow measurements .............................................................................. 44
2.2.2.1 Measurement of heat pulse velocities ................................................... 44
2.2.2.2 Wounding correction ............................................................................. 47
2.2.2.3 Determining sap velocity ....................................................................... 47
2.2.2.4 Converting sap velocity to sap flux density............................................ 49
2.2.2.5 Xylem anatomical assessments ............................................................ 50
2.2.3 Measurement of orchard evapotranspiration ............................................... 51
2.2.3.1 The eddy covariance system ................................................................. 51
2.2.4 Calibration of the HRM sap flow technique .................................................. 52
2.3 Results ............................................................................................................... 54
2.3.1 Calibration of the HRM sap flow technique .................................................. 54
2.3.3 Xylem anatomy ............................................................................................ 57
2.4 Discussion .......................................................................................................... 60
2.5 Conclusions ........................................................................................................ 62
Chapter 3: Measurement of long-term transpiration and transpiration coefficients in
citrus orchards .......................................................................................................... 63
3.1 Introduction ........................................................................................................ 63
3.2 Materials and methods ....................................................................................... 64
3.2.1 Experimental site and climate ...................................................................... 64
3.2.2 Estimation of transpiration ........................................................................... 66
3.2.3 Reference evapotranspiration ...................................................................... 68
© University of Pretoria
viii
3.2.4 Determination of transpiration (Kt) crop coefficients..................................... 69
3.3 Results and discussion ....................................................................................... 70
3.3.1 Measurement of long-term transpiration ...................................................... 70
3.3.2 Determination of transpiration crop coefficients (Kt)..................................... 74
3.4 Conclusions ........................................................................................................ 78
Chapter 4: Estimation of transpiration crop coefficients from ground cover and height
in citrus orchards ...................................................................................................... 80
4.1 Introduction ........................................................................................................ 80
4.2 Materials and Methods ....................................................................................... 81
4.2.1 Model description ......................................................................................... 81
4.2.2 Model calibration .......................................................................................... 83
4.2.3 Model validation ........................................................................................... 84
4.3 Results and discussion ....................................................................................... 84
4.4 Conclusions ........................................................................................................ 93
Chapter 5: Predicting transpiration of citrus orchards using the Penman-Monteith
equation coupled with a Jarvis-type multiplication canopy conductance model ....... 94
5.1 Introduction ........................................................................................................ 94
5.2 Materials and methods ....................................................................................... 95
5.2.1 Calculation of canopy conductance ............................................................. 95
5.2.2 Modelling canopy conductance.................................................................... 96
5.2.2.1 Model parameterisation ......................................................................... 97
5.2.2.2 Model validation .................................................................................... 98
5.2.3 Statistical analysis ....................................................................................... 98
5.3 Results ............................................................................................................... 99
5.3.1 Canopy conductance ................................................................................... 99
5.3.2 Parameterisation of the Jarvis model ......................................................... 102
5.3.2 Predicting transpiration using the Penman-Monteith equation coupled with
the Jarvis-type model .......................................................................................... 104
5.4 Discussion ........................................................................................................ 117
5.5 Conclusion ....................................................................................................... 119
Chapter 6: General conclusions and recommendations ......................................... 120
6.1 Overview of study ............................................................................................. 120
6.2 Calibration of the heat ratio method (HRM) sap flow technique for long-term
transpiration measurement in citrus orchards ........................................................ 121
© University of Pretoria
ix
6.3 Measurement of long-term transpiration and transpiration coefficients in citrus
orchards ................................................................................................................. 121
6.4 Estimation of transpiration coefficients from plant height and canopy cover .... 123
6.5 Predicting transpiration of citrus orchards using the Penman-Monteith equation
coupled with a Jarvis-type canopy conductance model.......................................... 123
6.6 Recommendations for future research ............................................................. 124
References ............................................................................................................. 126
© University of Pretoria
x
List of Figures
Figure 1. Citrus production according to province and type in South Africa in the
2014/15 growing season (Citrus Growers Association of Southern Africa,
2015a). .................................................................................................... 2
Figure 2. Quantity of oranges produced for local consumption, exports and
processing in South Africa (Source: Citrus Growers Association of
Southern Africa, 2015b). ......................................................................... 3
Figure 3. Gross values of oranges on the different markets in South Africa (Source:
Citrus Growers Association of Southern Africa, 2015b). ......................... 3
Figure 1.1 Schematic diagram of a probe set used for sap flow measurements
using the HRM showing the heater probe, upper (down-stream) and
lower (up-stream) temperature sensors. The depth of probe insertion d;
and the distance between the heater probe and both temperature
sensors x are also shown. ..................................................................... 17
Figure 1.2. Simplified representation of the (bulk) surface and aerodynamic
resistances for water flow through the soil-plant-atmosphere continuum
(Allen et al., 1998). ................................................................................ 24
Figure 1.3. Characteristics of the hypothetical reference crop used for calculation of
ETo in the FAO Penman-Monteith equation (Allen et al., 1998). ........... 26
Figure 2.1. Sap flow probes installed in two trees in the ‘Delta’ Valencia orchard.
Insert shows situation of the probes around a stem of one of the sample
trees. ..................................................................................................... 46
Figure 2.2. Sample of wood taken from one of the sample trees for the
determination of wounding width, sapwood moisture content and
sapwood density: (A) the sample of wood just before removal (B)
measurement of the wounding width and (C) the wood sample used for
measurements ...................................................................................... 48
Figure 2.3. Measurement of the wounding width on a wood sample cut transversely
from the trunk of one of the sample trees at the end of the measuring
season in August 2009 (photo courtesy of Mark Gush)......................... 48
Figure 2.4. Idealised cross-sectional areas of four annuli rings represented by probe
sets inserted at different depths in the conducting sapwood of a tree
stem shaded in different colours. .......................................................... 50
© University of Pretoria
xi
Figure 2.5. Micro-meteorological instruments including mounted on a lattice mast at
the centre of the ‘Delta’ Valencia orchard (photo courtesy of Mark Gush).
.............................................................................................................. 53
Figure 2. 6. Error analysis illustrating resultant percentage error in transpiration from
the HRM due to errors in selected parameters in the up-scaling
equations from Vh’s to transpiration, whilst maintaining all other
parameters at their original value. ......................................................... 55
Figure 2.7. Comparison between daily transpiration obtained using a virtual wound
width of 3.3 mm for the winter period and 3.1 mm for the autumn period.
.............................................................................................................. 57
Figure 2.8. Up-scaled HRM-based transpiration using a “virtual wound width” of 3.2
mm as compared to micrometeorological measurements of
evapotranspiration. ................................................................................ 58
Figure 2.9. Photomicrographs of sapwood samples taken from a ‘Valencia’ orange
tree. (A) A transverse section of the citrus sapwood prior to probe
insertion showing the distribution of the xylem vessels (x) in the
sapwood; (B) a transverse section of sapwood adjacent to one of the
drilled holes made for insertion of a probe; (C) tangential section
adjacent to one of the drilled holes showing the affected rays; (D) a
transverse section showing the extent of blockage of xylem vessels by
gum (g) and phenols (p); and (E) a transverse section of the sapwood
approximately 1 mm above the downstream thermocouple showing
extensive xylem blockage. Scale bars on each photograph are equal to 2
mm. ....................................................................................................... 59
Figure 3.1. Relationship between trunk circumference and seasonal water use for
the ‘Bahianinha’ and ‘Rustenburg’ Navel orchards ................................ 68
Figure 3.2. Daily transpiration and reference evapotranspiration determined in the
‘Delta’ Valencia, ‘Bahianinha’ and ‘Rustenburg’ Navel orchards. .......... 71
Figure 3.3. Transpiration coefficients (Kt) determined for the ‘Delta’ Valencia,
‘Bahianinha’ Navel and ‘Rustenburg’ Navel orchards............................ 75
Figure 3.4. Average monthly transpiration coefficients (Kt) for the three orchards.
Also shown are the FAO-56 standardised basal crop coefficients (Kcb
FAO-56) for a citrus orchard with 70 % canopy cover (‘Delta’ and
© University of Pretoria
xii
‘Rustenburg’ orchards) and 50 % ground cover (‘Bahianinha’ orchard),
all with no active ground cover, as given by Allen and Pereira (2009). . 76
Figure 4.1. Derived monthly transpiration crop coefficients (Kt) for the three
orchards. Transpiration crop coefficients were determined using
measured transpiration (Kt measured), the method described in Allen and
Pereira (2009) using the parameters given for citrus (Kt Allen and Pereira),
using estimates of monthly average rl values from transpiration data (Kt
monthly avg. rl) and using rl estimated from the relationship with ETo (Kt
rl (ETo)). Also shown are FAO-56 standardised basal crop coefficients
(Kcb FAO-56) for a citrus orchard with 70% canopy cover (‘Delta’ and
‘Rustenburg’ orchards) and 50% ground cover (‘Bahianinha’ orchard), all
with no active ground cover, as given by Allen and Pereira (2009) ....... 85
Figure 4.2. Monthly mean leaf resistances calculated following the procedure
outlined by Allen and Pereira (2009), compared with the values
suggested by Allen and Pereira (2009) for citrus and daily stomatal
resistance measure in the ‘Rustenburg’ Navel orchard in Citrusdal. ..... 86
Figure 4.3. Relationship between reference evapotranspiration (ETo) and mean leaf
resistance for the ‘Rustenburg’ Navel orchard ...................................... 88
Figure 4.4. Monthly measured and estimated transpiration using Kt values derived
from mean leaf resistance, estimated from the relationship between
reference evapotranspiration and mean leaf resistance in the three
orchards. Statistical parameters shown include mean absolute error
(MAE), root of the mean square error (RMSE) and index of agreement
(D) ......................................................................................................... 89
Figure 4.5. Cumulative ETo, measured transpiration and transpiration estimated
from the derived crop coefficients in the ‘Delta’ Valencia orange orchard
(1 August 2008 to 30 July 2009), the ‘Bahianinha’ Navel orchard (3
September 2009 to 30 June 2010) and the ‘Rustenburg Navel orchard
(13 August 2010 to 12 August 2011) ..................................................... 92
Figure 5.1. Typical diurnal trend of A) solar radiation (SR), B) temperature (T), C)
vapour pressure deficit (VPD) and D) canopy conductance (gc),
calculated from the inversion of the Penman-Monteith, in the ‘Delta’
Valencia orchard on a clear sky day, 27 January 2010. ...................... 100
© University of Pretoria
xiii
Figure 5.2. Hourly canopy conductance (gc) calculated using the measured
transpiration and inverting the Penman-Monteith equation in the three
orchards. Gaps in the plotted data were due to missing hourly weather
data caused by power failure to the weather stations.......................... 101
Figure 5.3. Normalised canopy conductance in the ‘Delta’ Valencia orchard plotted
against (A) solar radiation (B) vapour pressure deficit and (C)
temperature. The forms of the model functions, based upon the
optimised parameters, are shown as solid lines. ................................. 103
Figure 5.4. Relationship between hourly gc predicted by the Jarvis model and gc
obtained from sap flow data by inverting the Penman-Monteith equation
in the ‘Delta’ Valencia orchard. ............................................................ 104
Figure 5.5. Measured and predicted daily transpiration in the ‘Delta’ Valencia
orchard for A) the period 16 December 2008 to 26 June 2009 and (B)
regression analysis. Statistical parameters include slope of regression
equation, coefficient of determination (R2), mean absolute error (MAE),
Willmot coefficient of agreement (D) and root mean square error
(RMSE). The missing data was due to missing hourly weather data
caused by power failure to the weather station. .................................. 105
Figure 5.6. Relationship between gc predicted by the Jarvis model using parameters
of the ‘Delta’ Valencia orchard and gc obtained from sap flow data by
inverting the Penman-Monteith equation in the ‘Bahianinha’ and
‘Rustenburg’ Navel orchards. .............................................................. 107
Figure 5.7. Time series of measured and predicted transpiration in the ‘Bahianinha’
and ‘Rustenburg’ Navel orchards using Jarvis parameters optimised in
the ‘Delta’ Valencia orchard. ............................................................... 108
Figure 5.8. Regression analysis between measured and predicted transpiration in
the ‘Bahianinha’ and ‘Rustenburg’ Navel orchard using parameters
determined in the ‘Delta’ Valencia orchard. Statistical parameters used
include slope of regression equation, coefficient of determination (R2),
mean absolute error (MAE), Willmot coefficient of agreement (D) and
root mean square error (RMSE). ......................................................... 109
Figure 5.9. Normalised canopy conductance in the ‘Bahianinha’ Navel orchard
plotted against (A) solar radiation (B) vapour pressure deficit and (C)
© University of Pretoria
xiv
temperature. The forms of the model functions based upon the optimised
parameters are shown as solid lines. .................................................. 111
Figure 5.10. Normalised canopy conductance in the ‘Rustenburg’ Navel orchard
plotted against (A) solar radiation, (B) vapour pressure deficit and (C)
temperature. The forms of the model functions based upon the optimised
parameters are shown as solid lines. .................................................. 112
Figure 5.11. Relationship between hourly gc predicted by the Jarvis model and gc
obtained from sap flow data by inverting the Penman-Monteith equation
in the ‘Bahianinha’ and ‘Rustenburg’ Navel orchards. ......................... 113
Figure 5.12. Time series plot of measured and predicted transpiration in the
‘Bahianinha’ and ‘Rustenburg’ Navel orchards. The gaps in the plotted
data were due to missing hourly weather data caused by power failure to
the weather stations. ........................................................................... 114
Figure 5.13. Regression analysis between predicted and measured transpiration in
the ‘Bahianinha’ and ‘Rustenburg’ Navel orchards. Statistical parameters
used include slope of regression equation, coefficient of determination
(R2), mean absolute error (MAE), Willmot coefficient of agreement (D)
and root mean square error (RMSE). .................................................. 115
Figure 5.14. Cumulative measured transpiration and transpiration predicted using a
Jarvis-type model parameterised in the respective orchards for the
measurement periods. The plateau in water use graphs occurring
around April/May in the three graphs is due to periods of missing data.
............................................................................................................ 116
© University of Pretoria
xv
List of Tables
Table 1.1. Average annual and seasonal crop water use (mm day-1) of citrus tree
orchards, in different growing areas of the world. .................................. 10
Table 1.2. Citrus crop coefficient (Kc and Kcb) values for standard climate of RHmin =
45 % and u2 = 2 m s-1 as published for use with FAO Penman-Monteith
ETo (Allen et al. 1998). .......................................................................... 28
Table 1.3. Seasonal crop coefficients determined in citrus orchards across the
growing regions of the world. ................................................................ 30
Table 1.4. Generic parameters for a citrus crop for estimating crop coefficients. .... 35
Table 1.5. Citrus values of Kcini, Kc mid, Kc end, Kcbini, Kcb mid and Kcb end for standard
climate of RHmin = 45 % and u2 = 2 m s-1 as expanded from FAO-56
(Allen et al. 1998) for a range of values for effective ground covered by
the tree canopy (fc eff) during the midseason and using the procedure of
Allen and Pereira (2009) with recommended parameter values. ........... 35
Table 2.1. Established circumference size classes (CSC) established, the number
of trees in the class (n), the sample tree representing each class, sample
tree circumference (Sc) and the fraction represented by each sample
tree in the ‘Delta’ Valencia orchard as a percentage (pt). ...................... 45
Table 2.2. Sapwood water content and sapwood density for various citrus species. A
total of 19 samples were taken, with six from lemon, nine from sweet
orange, two from soft citrus and two from grapefruit. ............................ 55
Table 2.3. Slopes and intercepts of the regression equations used to determine
wounding width by inputting ETc on different days. The R2 values and the
virtual wound widths determined from the regression equations are also
shown. ................................................................................................... 56
Table 3.1. Established circumference size classes (CSC), the number of trees in the
class (N), the sample tree representing each class, sample tree
circumference (Sc) and the percentage represented by each sample tree
each orchard (Pt). .................................................................................. 67
Table 3.2. Average daily transpiration (T) and reference evapotranspiration (ETo) for
the summer and winter seasons in the ‘Delta’ Valencia and ‘Bahianinha’
Navel. .................................................................................................... 72
© University of Pretoria
xvi
Table 3.3. Average transpiration (T), reference evapotranspiration (ETo) and
transpiration crop coefficients (Kt) for the summer and winter seasons. 77
Table 5.1. Optimised parameters for the Jarvis model used to predict canopy
conductance with vapour pressure deficit, solar radiation and ambient
temperature as response functions in the ‘Delta’ Valencia orchard. .... 102
Table 5.2. Parameters for the response functions of vapour pressure deficit, solar
radiation and temperature for the Jarvis model determined through the
optimisation process in the ‘Bahianinha’ and ‘Rustenburg’ Navel
orchards. ............................................................................................. 110
© University of Pretoria
xvii
List of abbreviations and symbols
Abbreviations
CHPM compensation heat pulse method
CSC circumference size classes
DWAF Department of Water Agriculture and Forestry
ETc orchard evapotranspiration
FAO Food and Agriculture Organization of the United Nations
HRM heat ratio method
IPCC Inter-Governmental Panel on Climate Change
IRGA open path infrared gas analyser
LAI leaf area index
MAE mean absolute error
OPEC open path eddy covariance
Sc sample tree circumference
TC thermocouple
Symbols
Symbol Description
Ab canopy basal area
AT total area allocated to each tree in the orchard
cw specific heat capacity of the wood matrix
cs specific heat capacity of the sap
d characteristic leaf dimension
ETc crop evapotranspiration
ETo reference evapotranspiration
ƒceff effective fraction of ground covered or shaded by vegetation
near solar noon
Fr adjustment factor relative to stomatal control
G Ground heat flux
gc canopy conductance
gc max maximum canopy conductance
h mean maximum plant height during the midseason period or
full cover period
© University of Pretoria
xviii
k thermal diffusivity of green (fresh) wood
Kcb basal crop coefficient
Kcb full basal crop coefficients for fully grown orchard
Kc min minimum crop coefficient for bare soil
Kd crop density coefficient
Ke soil water evaporation coefficient
K canopy extinction coefficient of light attenuation
Mw wet mass of the sapwood sample
Md dry mass of the sapwood sample
ML parameter that simulates the physical limits imposed on water
flux through the plant root, stem and leaf systems
N number of sample trees on which transpiration was conducted
pti portion of trees represented by the ith tree sampled in the
orchard
Q sap flux
Rw inter-row spacing
Reff effective precipitation
RHmin minimum relative humidity
rk radii forming the annuli rings of the sap wood measured from
the centre of the trunk
rl mean leaf resistance
Rn net irradiance
Rn total net radiation flux density
SR solar radiation
RO run-off
rs estimated mean canopy bulk resistance for the citrus trees
T orchard transpiration
Ta mean daily air temperature at 2 m height
TL lower temperature limit to transpiration
TU upper temperature limit to transpiration
Tpost temperature measured after the heat pulse
Tpre average temperature before the heat pulse
u2 wind speed at 2 m height
© University of Pretoria
xix
v1 increase in temperature of upper thermocouple after the heat
pulse is released
v2 increase in temperature of the lower thermocouple after the
heat pulse is released
Vc corrected heat pulse velocity
Vh heat pulse velocity
VPD vapour pressure deficit
Vs sap velocity
Vw volume of wood sample
W average or effective tree canopy width
w wounding width
Δ slope of the vapour pressure curve
δM function of soil water content deficit
ρb basic density of wood
ρs density of water
γ psychrometric constant
© University of Pretoria
1
Introduction and structure of thesis
The genus Citrus (family Rutaceae) is a range of fruits which includes oranges,
mandarins, tangerines, tangelos, clementines, satsumas, lemons, limes, and
grapefruits (Syvertsen and Lloyd, 1994; Carr, 2012). Citrus are perennial evergreen
trees believed to have originated from the humid tropics of the south-eastern parts of
China and in southern India (Kriedemann and Barrs, 1981; Sippel, 2006). The
species have, however, become widely adapted to the semi-arid regions of the world
(Carr, 2012). They are cultivated primarily for fresh fruit and juice, although there are
other by-products, such as food additives, pectin, marmalades, cattle feeds (from the
peel), cosmetics, essential oils, chemicals, and medicines.
Citrus is a one of the most important fruit crops grown across the world, as well as in
South Africa. South Africa has an average production of 2.16 million tons per year,
which accounts for 4.2 % of world production and as a result the country is ranked
thirteenth in the world in terms of citrus production (FAO, 2012). In South Africa,
Valencia (including mid-seasons) and Navel orange orchards occupy the bulk of the
64 510 ha planted with citrus at the present moment, accounting for 40 and 24.3 %
of the total area planted to citrus, respectively (Citrus Growers Association of
Southern Africa, 2015a) (Figure 1). Figure 2 shows total orange production and the
quantity of fruit that was used for local consumption, processed and exported in
South Africa in recent years (Citrus Growers Association of Southern Africa, 2015b).
While the quantity of fruit used for local consumption and processing has remained
fairly constant over the years, there has been a gradual increase in the fruit sold on
the export market. This is probably due to the ever increasing price of fresh fruit on
the export markets (Figure 3) where South Africa is ranked third.
Citrus is grown in all provinces of South Africa with rare to frost-free conditions (this
excludes Gauteng and Free State) (Bijzet, 2006). The major producing provinces are
the Eastern Cape, Mpumalanga, Limpopo and Western Cape (Figure 1). A large
portion of these areas receive seasonal rainfall, less than 500 mm per annum,
making them semi-arid and therefore irrigation is required in order to meet crop
water requirements and enable the successful production of citrus in these areas.
© University of Pretoria
2
Figure 1. Citrus production according to province and type in South Africa in the
2014/15 growing season (Citrus Growers Association of Southern Africa,
2015a).
© University of Pretoria
3
Year
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Qu
antity
of o
ran
ge
s (
t ye
ar-1
)
0.0
2.0e+5
4.0e+5
6.0e+5
8.0e+5
1.0e+6
1.2e+6
1.4e+6
1.6e+6
1.8e+6
2.0e+6
Total
Local
Export
Processed
Figure 2. Quantity of oranges produced for local consumption, exports and
processing in South Africa (Source: Citrus Growers Association of
Southern Africa, 2015b).
Year
2004 2006 2008 2010 2012 2014
Gro
ss v
alu
e o
f ora
nges (
Rands t
-1)
0
1000
2000
3000
4000
5000
6000
7000
Average
Exports
Processed
Figure 3. Gross values of oranges on the different markets in South Africa (Source:
Citrus Growers Association of Southern Africa, 2015b).
© University of Pretoria
4
Currently, it is estimated that irrigated agriculture uses approximately 70 % of the
available fresh water resources on a global scale (FAO-AQUASTAT, 2003) and in
South Africa it is estimated at 60 % (DWAF, 2013). It has been predicted that as the
world population expands and economies grow, competition for water will increase,
while water resources become more polluted. These changes are expected to be
exacerbated as a result of climate change, which is believed to have already
increased the frequency and severity of droughts in southern Africa in recent years
(IPCC, 2007). Consequently, the demand for water will exceed supply, and less
water will be available for agriculture. These population and climate dynamics will
have catastrophic effects on 90 % of the fruit production industry in South Africa
which, according to Annandale et al. (2005), is entirely dependent on irrigation water.
Water-saving agricultural practices and sound water management strategies are
therefore urgently required to ensure the long-term sustainability of the industry.
The advent of precision irrigation systems, such as drip, provides an opportunity to
match crop water requirements and irrigation amounts, but needs to be coupled with
appropriate water management techniques (Jones, 2004). Insufficient knowledge of
fruit tree water use under these relatively new systems has been identified as a
major factor hindering the development of sound water management tools in South
Africa (Pavel et al., 2003; Volschenk et al., 2003). Available literature on citrus water
use measurements is in the form of Green and Moreshet (1979) and Du Plessis
(1985), which emphasizes that very little has been done in terms of research on
citrus water use in South Africa in recent years. There is need to update such
information to cater for changes in production practices.
Accurate and reliable methods of measuring water use in fruit tree orchards are
needed to gather information to fill this knowledge gap. Crop evapotranspiration
(ETc) can be measured using a number of systems which include lysimeters, micro-
meteorological methods, soil water balance, sap flow, scintillometry and satellite-
based remote sensing (Allen et al., 2011). Quantification of the transpiration
component that forms the larger portion of orchard water use (Reinders et al., 2010)
and is directly related to productivity (Villalobos et al., 2013), is critical in micro-
irrigated fruit tree orchards. Reviews by Swanson (1994), Smith and Allen (1996),
Wullschleger et al. (1998) and Allen et al. (2011) have reported sap flow
© University of Pretoria
5
measurements as accurate and reliable methods for estimating long-term
transpiration of woody species such as citrus trees, provided the method is
calibrated and that due care is taken during installation of the equipment.
Measurements of crop water use are considered to be expensive and time
consuming, requiring specialised expertise, which is not always available. Physically-
based models can be used for the extrapolation of measured crop water use to
different environments for the improvement of irrigation water management (Boote et
al., 1996). The term irrigation water management used here encompasses the
activities of irrigation system planning, irrigation scheduling and issuing of water
rights/permits. According to Allen et al. (2011), ETc is typically modelled on a
physical basis using weather data and algorithms that describe the surface energy
and aerodynamic characteristics of the crop. Simultaneous measurements of ETc
and weather data are required to improve the current models; as well as adapt the
models to account for new production practices e.g. use micro-irrigation.
The crop coefficient approach first published by Doorenbos and Pruitt (1977), and
later refined by Allen et al. (1998), has been used extensively and is currently
considered as the standard model of ETc. Although site specific crop coefficients are
probably best, the crop coefficients for a generic citrus tree orchard are often used
for water management, due to a lack of reliable information on tree water use. The
basal crop coefficient (Kcb), which is directly linked to transpiration, is very important
in micro-irrigated orchards, as it represents the major component of orchard water
use and is directly linked to productivity (Villalobos et al., 2013). Villalobos et al.
(2013) highlighted that the Kcb factors published by Allen et al. (1998) contain some
residual diffusive evaporation component supplied by soil water below the dry
surface and by soil water from beneath dense vegetation; such that the term
transpiration coefficients (Kt) should be used when basal crop coefficients are
determined using transpiration determined by sap flow methods.
The crop coefficient approach is often used in irrigation water management because
it is relatively simple to implement. Research has shown much variation in crop
coefficients reported for citrus orchards across the different growing regions of the
© University of Pretoria
6
world (García Petillo and Castel, 2007; Snyder and O’Connell, 2007). This means
that crop coefficients may be site specific and may not be readily transferable from
one orchard to another (Testi et al., 2006). Allen and Pereirra (2009) consolidated
the FAO56 procedure for estimating crop coefficients and developed a method for
developing site specific crop coefficients based on the fractional cover and height of
the vegetation and the degree of stomatal control over transpiration relative to most
agricultural crops. The use of this approach still needs to be tested against Kt
determined using sap flow measurements in citrus orchards.
Citrus trees have been associated with high resistances that play a major role in the
regulation of water movement within the plant (van Bavel et al. 1967; Kriedemann
and Bars, 1981; Meyer and Green, 1981; Sinclair and Allen, 1982). Sap flow-based
transpiration measurements have proved useful in the parameterisation of canopy
conductance models for different tree species including citrus (e.g. Rana et al., 2005;
Oguntunde et al., 2007; Villalobos et al., 2009; Villalobos et al., 2013). A canopy
conductance model, that has been evaluated for short periods, as in the case for the
studies for citrus cited above, is the multiplicative approach that was conceived by
Jarvis (1976). Besides aiding scientists in understanding the bulk behaviour of
stomatal conductance with respect to changing environmental conditions
(Wullschleger, 1998), it has also been incorporated into the Penman-Monteith
equation to estimate transpiration. The use of such a canopy conductance modelling
approach can play a vital role in mechanistically modelling transpiration of citrus
trees, where hydraulic conductance plays an important role in regulating
transpiration.
This study was part of a project solicited, initiated, managed and funded by South
Africa’s Water Research Commission (Project K5/1770) entitled “Water use of fruit
tree orchards” with core objectives of measuring and modelling of water use of
selected fruit tree orchards.
© University of Pretoria
7
Aim of the study
The aim of this study was to measure transpiration of well managed micro-irrigated
citrus orchards and test physically–based models that can be used to estimate
transpiration across different citrus growing regions.
Thesis objectives
The objectives of the thesis were to:
1. Calibrate the heat ratio method (HRM) sap flow technique using micro-
meteorological measurements for long-term transpiration measurement in a
citrus orchard
2. Quantify long-term transpiration and transpiration crop coefficients in citrus
orchards using the calibrated sap flow technique
3. Evaluate the appropriateness of the canopy cover-based transpiration crop
coefficient approach as a method for estimating transpiration in citrus
orchards
4. Use sap flow-based transpiration measurements to parameterise a Jarvis-
type conductance model for inclusion in the Penman-Monteith model for
predicting long-term transpiration of citrus orchards
Hypotheses
The main hypotheses that were formulated and tested in this study were as follows:
Citrus sapwood matrix is not thermally homogeneous such that conversion of
measured heat pulse velocity to transpiration requires an empirical calibration
factor which will be constant for long-term measurements
A dynamic estimate of leaf (or canopy) resistance is needed to predict the
daily and seasonal changes of transpiration in citrus orchards
Thesis outline
A review of the current knowledge on the measurement of evapotranspiration in
citrus is initially presented in Chapter 1. The main body of the thesis consists of four
chapters. The first two chapters (2 and 3) cover the specifics of measurement of
© University of Pretoria
8
water use in three citrus orchards. Chapter 2 covers the calibration of the heat ratio
method (HRM) using micrometeorological measurements for estimating sap flow for
transpiration measurements in a ‘Delta’ Valencia orchard. In Chapter 3 details of
measuring long term transpiration using the calibrated HRM in three citrus orchards
are presented. Transpiration coefficients were determined for the three orchards
based on measured transpiration data and reference evapotranspiration determined
using the FAO56 Penman-Monteith equation. Physically-based models that can be
used for predicting transpiration of citrus orchards were tested in Chapters 4 and 5.
In Chapter 4, crop coefficients determined from measurements of canopy cover and
tree height, following the method of Allen and Pereira (2009), were evaluated against
those calculated based on measured transpiration and reference evapotranspiration
data. In Chapter 5, a mechanistic approach of modelling long-term citrus
transpiration using the Penman-Monteith equation coupled with a Jarvis-type canopy
conductance model was tested. Chapter 6 gives the general conclusions reached in
this work and some recommendations for future studies in relation to measurement
and modelling transpiration in citrus orchards.
© University of Pretoria
9
Chapter 1: Literature Review
1.1 Measurement of citrus water use
1.1.1 Introduction
Knowledge of crop evapotranspiration (ETc) is the fundamental basis for efficient
irrigation water management strategies. The term irrigation water management
encompasses activities such as irrigation systems planning, irrigation scheduling and
issuing of water rights/permits. Besides irrigation water management knowledge of
crop evapotranspiration is also used across various disciplines as explained by Allen
et al. (2011) “crop ETc information is more and more frequently used as a foundation
for court determinations of injury among water users, for parameterization of
important hydrologic and water resources planning and operation models, for
operating weather and climate change forecasting models, and for water
management and allocation in water-scarce regions, including the partitioning of
water resources among states and nations”.
A great deal of research has been published on the water relations of Citrus spp.
Several reviews that synthesize this information have been published, which include
photosynthesis (Ribeiro and Machado, 2007), physiological aspects of the control of
the water status (Jones, 1985), irrigation (Shalhevet and Levy, 1990), physiological
responses to the environment (Syvertsen and Lloyd, 1994) and general water use
(Kriedemann and Barrs, 1981; Carr, 2012). Published data for ETc of Citrus species
across the different growing regions of the world is presented in Table 1.1. According
to these studies, citrus water use ranges from 0.8-8.5 mm day-1. Variation of crop
water use in citrus orchards can be attributed to a number of factors that include
varieties grown, rootstock, tree spacing, canopy height, ground cover, tillage,
microclimate, irrigation method and frequency; and method of measuring crop water
use (Snyder and O’Connell, 2006; Naor et al., 2008).
© University of Pretoria
10
Table 1.1. Average annual and seasonal crop water use (mm day-1) of citrus tree orchards, in different growing areas of the world.
Reference Location Tec. CM Winter Spring Summer Autumn Annual Age spha
Koo and Sites (1955)* Florida, USA SWB ETc 1.7 2.7 4.0 2.8 2.8
Kalma and Stanhill (1969)* (Shamouti oranges)
Israel SWB ETc 1.1 4.4 2.3
Green and Moreshet (1979) (Valencia oranges)
South Africa Lys ETc 2.7 2.2 4.6 5.2 3.68 15
Hoffman et al. (1982) (Oranges)
Arizona, USA SWB ETc 1.1 4.0 7.3 3.7 4.0
Smajstrla et al. (1982) Florida, USA SWB ETc 1.9 3.9
Wiegand and Swanson (1982) (Valencia oranges)
Texas, USA SWB ETc 1.8 2.6 4.8 3.7 3.2
Rogers et al. (1983) Florida, USA SWB ETc 5 3.3 8
Du Plessis (1985) (Valencia oranges)
South Africa Lys ETc 2.3 6.0-8.5
Castel et al. (1987)* (Oranges)
Valencia, Spain SWB ETc 0.9-1.0 1.5-2.0 3.0-3.2 1.4-1.8 1.7-2.0
Sepaskhah and Kashefipour (1995)* (Lemons)
Iran SWB ETc 1.4 4.4 8.5 4.2 4.6
Castel (1997) (Clementines)
Spain Lys ETc 0.5 1.5 1 7-11 432
Fares and Alva (1999) (Hamlin oranges)
Florida,USA SWB ETc <1.0 >4.0
*Water use figures obtained from other sources and not the original reference Tec – technology used to measure water use: Lys – lysimeter; SWB – soil water balance; SF – sap flow; SR – surface renewal; EC – eddy covariance; μLys – microlysimeter CM – component of crop evapotranspiration that was measured: ETc –crop evapotranspiration; T – transpiration; Es – soil evaporation spha – stems per hectare (i.e. number of trees per hectare); Table 1.1 continued
© University of Pretoria
11
Reference Location Tec. CM Winter Spring Summer Autumn Year Age spha
Castel (2000) (Clementines)
Valencia, Spain Lys ETc 0.7-1.0 2.0-2.8 1.9
Yang et al. (2003) (Murcott oranges)
Japan, Greenhouse
Lys ETc 0.6 5 4.4 8 4444
Rana et al. (2005) (Clementines)
Italy SF T 3 8 4 10 400
Consoli et al. (2006) (Navel oranges)
California, USA SR ETc 0.7mm h-
1 July-
August 4 337
0.8mm h-
1 15 299
0.9mm h-
1 34-
36 284
Alarcon et al. (2006) (Lemons)
Spain SF
T 1.8L day-
1 1.4 L day-1 2
García Petillo and Castel (2007) (Valencia oranges)
San Jose, Uruguay
SWB ETc 1.3 2.8 3.0 1.1 2.1
Snyder and O’Connell (2007) (Navel oranges)
Linsay, California SR ETc 2.36 3.86 5.55 4.7 4.14 33-37
283
Villalobos et al (2009) (Clementina mandarin)
EC μLys
T Es
1.3 0.82
1.8 0.82
17 333
Marin and Angelocci (2011) (Lemons)
Sao Paulo, Brazil SF T 0.8 2 7 179
ETc 0.8 2.7
Villallobos et al. (2013) (Navel oranges)
Southern Spain SF T 1.3 2.3 11 476
0.7 1.5 10 416
© University of Pretoria
12
Table 1.1 also shows that the soil water balance has been used quite extensively to
measure ETc (and/or its components) of citrus. Other methods that have been used
for determining citrus ETc and/or its components include the micro-meteorological
methods (eddy covariance), lysimeters, and sap flow. Carr (2012) and García Petillo
and Castel (2007) highlighted that it is not easy to quantify the water requirements of
an orchard crop such as citrus or to compare the results of measurements taken
under different conditions and with different methods. Generally ETc measurements
are considered to be expensive, time consuming and requiring specialised expertise
(Allen et al., 2011). The situation is made worse in fruit tree orchards due to the
variability of tree sizes and the perennial nature of the crop, which means long time
periods are required to gather such information (Hoffman et al., 1982; Castel et al.,
1987).
Such dynamics may have had an effect on research of ETc of fruit trees in South
Africa, where such information is considered lacking (Pavel et al., 2003; Volschenk
et al., 2003). Research work from South Africa cited in Table 1.1 and in other
reviews (e.g. Carr, 2012; García Petillo and Castel, 2007) pertaining to citrus water
use was performed over three decades ago. There is need to update such
information to cater for changes in production practices such as rootstock choice
(from the more vigorous ‘Rough Lemon’ to the rootstocks of intermediate vigour
which include ‘Carrizo’, ‘Troyer’ and ‘Swingle’); and adoption of micro-irrigation
systems.
The use of micro-irrigation systems helps to maximise transpiration (i.e. beneficial
water use) and minimise evaporation (i.e. non-beneficial water use) in farming
systems (Reinders et al., 2010). It is important that ETc measurements in micro-
irrigated orchards focus on the transpiration component as it represents the major
component of ETc and is directly linked to productivity (Villalobos et al., 2013).
Thermometric sap flow techniques have shown great promise as methods for
measuring transpiration of woody tree species such as citrus (Smith and Allen, 1996;
Wullschleger et al., 1998). Sap flow methods have recently been used to quantify
transpiration of citrus orchards on a long-term basis by Marin and Angelocci (2011)
and Villallobos et al. (2013). Such methodologies can be utilised to expedite the
gathering of information on transpiration of citrus orchards in South Africa.
© University of Pretoria
13
The next section reviews thermometric sap flow methods of measuring transpiration
in woody species. Emphasis is placed on the operational aspects that need to be
considered when quantifying long-term transpiration in orchards.
1.1.2 Sap flow measurements of crop water use
A number of methods have been used to directly estimate transpiration of trees in
general, including fruit trees (Wullschleger et al., 1998). These include weighing
lysimeters, large tree potometers, ventilated chambers, radioisotopes, stable
isotopes and a range of heat balance/heat dissipation methods. The most accurate
and accepted method is the weighing lysimeter and as such it has been used in a
number of studies in orchards (Moreshet and Green, 1984; du Plessis, 1985; Castel,
1996; Girona et al., 2002; Ayars et al., 2003; Girona et al., 2003; Yang et al., 2003;
Williams and Ayars, 2005). However, they are expensive to install in existing
orchards, as well as to maintain.
Amongst the several methods that have been used to measure tree transpiration, the
use of thermometric sap flow measurements has shown considerable promise
(Smith and Allen, 1996; Wullschleger et al., 1998; Allen et al., 2011). Sap flow
measurements at stem level can be used to estimate tree transpiration if performed
over sufficiently long periods to negate changes in stem storage and neglecting the
2-5 % of the water used for photosynthesis (Salisbury and Ross, 1978; Waring and
Roberts, 1978). Some of the advantages of these methods include their ability to
estimate transpiration, which has a direct relationship with productivity; easy
interface to data loggers for direct automated transpiration measurements; their
applicability is not inhibited by complex terrain and heterogeneity within fruit tree
orchards; and they exhibit relatively good accuracy when compared to conventional
methods of measuring orchard water use, e.g. the eddy covariance method (Rana et
al., 2005).
Thermometric techniques that have been used for transpiration measurements in
woody species can be broadly categorized into heat pulse, heat balance and heat
dissipation methods (Smith and Allen, 1996). Heat pulse techniques, particularly the
compensation heat pulse method (CHPM), have been widely used, mainly because
© University of Pretoria
14
they are relatively easy to use and have low power requirements (Green et al., 2003;
Edwards et al., 1997; Smith and Allen, 1996). However, there are limitations to the
CHPM, of which the major shortcoming is the inability to measure low and reverse
sap flow rates, which usually occur under water stress conditions or at night (Becker,
1998). Burgess et al. (2001) developed the heat ratio method (HRM) as an
improvement to the CHPM, in order to cater for low and reverse sap flow rates in
woody species. The HRM has been used to study transpiration dynamics in Jatropha
curcas L. (Gush, 2008), olive (Williams et al., 2004; Er-Raki et al., 2010), Eucalyptus
marginata (Bleby et al., 2004) and Prosopis (Dzikiti et al., 2013).
1.1.2.1 Theory of heat pulse sap flow technique
The idea of using heat as a tracer of sap flow in woody tree species has been
perfected by several scientists over the years (Swanson, 1994). Marshall (1958)
further developed the theoretical framework used for heat pulse techniques, based
on a set of analytical solutions to the following heat flow equation:
𝜌𝑤𝑐𝑤𝑑𝑇
𝑑𝑡=
𝑑
𝑑𝑥𝜆𝑥
𝑑𝑇
𝑑𝑥+
𝑑
𝑑𝑦𝜆𝑦 − 𝑎𝑢𝜌𝑠𝑐𝑠
𝑑𝑇
𝑑𝑥+ 𝑄 Equation 1.1
where dT is the temperature departure from ambient (K) over an axial distance dx ,
dt is an increment in time (s), λ is the thermal conductivity (W m-1 K-1) in the axial (x)
and tangential (y) directions, a is the fraction of xylem cross sectional area occupied
by sap streams moving with a velocity u in the x-direction, and Q is the amount of
internal heat flux per unit length that is released from the heater (W m-3). Equation
1.1 ideally represents the two dimensional pattern of temperature surrounding a line
heater of zero dimension that is inserted into a section of sapwood of uniform
physical and thermal properties.
The velocity of the heat pulse carried by the moving sap stream is used to estimate
the rate of sap flow. Following the application of a heat pulse, the temperature rise in
the sapwood (Ts) at a distance from the heater is given by Marshall (1958) as:
𝑇𝑠 =𝑄
4𝜋𝜅𝑡𝑒𝑥𝑝 [
(𝑥−𝑉ℎ)+𝑦2
4𝜅𝑡] Equation 1. 2
© University of Pretoria
15
where κ is the thermal diffusivity (m2 s-1) of the sapwood, t is time since the heat
pulse was applied, (x,y) is the point in the xylem at which temperature
measurements are made at time t and Vh is the heat pulse velocity.
Measurements of Vh are effected by inserting a line heater and a set of temperature
sensors into the sapwood of the species to be measured. In order to ensure that the
heater and temperature sensors are correctly aligned a guide jig is normally used
when drilling the holes. At least four probes consisting of the heater probe and
temperature sensors are required to capture the radial variations of sap flow rates in
the sapwood (Hatton et al., 1990). The method is suitable only for use on woody
stems of at least 30 mm in diameter that can accommodate the probes (Smith and
Allen, 1996). Its use is also limited in large stems where the entire sapwood cannot
be accessed.
1.1.2.2 Compensation heat pulse method
Of the heat pulse techniques the CHPM has been the most widely used (Goodwin et
al., 2006; Edwards et al., 1996). In the CHPM a heater probe is inserted radially into
the sapwood portion of the stem together with two temperature sensors located up-
and down-stream of the heater probe (commonly –5 and 10 mm, respectively). To
monitor the sap velocity, wood and sap are heated in pulses and heat is carried
through convection by the moving sap stream. When both temperature sensors have
warmed to the same temperature, the heat pulse would have moved to the midpoint
between the probes. The heat pulse velocity (Vh) is calculated from the time taken
for the heat pulse to cover this distance as follows:
𝑉ℎ =𝑥1+𝑥2
2𝑡0× 3600 Equation 1.3
where to is time (s) for the downstream and upstream temperature probes to reach
thermal equilibrium after the heat pulse has been released; and x1 and x2 denote
distance in mm between the heater and the downstream and upstream temperature
probes respectively. A negative value is assigned to x2 because it is located on the
opposite side of the heater to x1.
© University of Pretoria
16
The principle used for measuring Vh in the CHPM has been found to overestimate
low, zero and reverse sap flow rates (Becker, 1998). This is because when sap flow
rates are low, the heat pulse may dissipate by conduction before it reaches the
measurement point, such that the sensors record equal temperatures (Burgess et
al., 2001). Lower thresholds for measurement of Vh depend on the sensitivity of
temperature sensors to capture the temperature changes, as well as the rate of
thermal diffusivity in xylem. Various lower limits that are undistinguishable from zero
have been reported e.g. 94 - 157 mm h–1 by Barrett et al. (1995); 36 - 72 mm h–1 by
Becker (1998); 40 mm h–1 by Burgess et al. (1998) and 30 mm h–1 by Swanson and
Whitfield (1981). Measured peak Vh in most woody species is seldom greater than
350 - 450 mm h–1, with mean rates less than half this value (Burgess et al., 2001).
Consequently, an inability of the CHPM to measure Vh below 30 - 40 mm h–1 will
affect the lower 10 - 30% of measurements. Measurements of transpiration during
winter, at night (Benyon, 1999) or on cloudy days are likely to be most affected. Such
erroneous measurements would have an impact on the long-term transpiration of
citrus which is known to have low conductance to water uptake.
1.1.2.3 Heat ratio method
Burgess et al. (2001) developed the HRM as an improvement to the CHPM to cater
for low and reverse sap flow rates in woody species. In the HRM, the ratio of the
increase in temperature is measured at points equidistant from a central heater soon
after a heat pulse is released. Heat pulse velocity (Vh) is calculated as:
2
13600v
vln
x
kVh Equation 1.4
where k is thermal diffusivity of green (fresh) wood (assigned a nominal value of 2.5
x 10-3 cm2 s-1), x is distance (cm) between the heater and either temperature probe,
and v1 and v2 are increases in temperature (from initial temperatures) at equidistant
points downstream and upstream (x cm) from the heater (Figure 1.1). During
operation temperature increases are measured and v1/v2 ratios are recorded
between 60 and 100 s after the release of the heat pulse. Vh are calculated based on
the average of the ratios during a measurement occasion of approximately 40 s.
Burgess et al. (1998; 2000) noted that there was a lesser occurrence of unexplained
variations of Vh data measured using the HRM as compared to CHPM equivalents
due to this improvement in sampling of measurements.
© University of Pretoria
17
Figure 1.1 Schematic diagram of a probe set used for sap flow measurements
using the HRM showing the heater probe, upper (down-stream) and
lower (up-stream) temperature sensors. The depth of probe insertion d;
and the distance between the heater probe and both temperature
sensors x are also shown.
1.1.2.4 Correction for wounding
Heat pulse techniques are invasive in nature because the probes are inserted in the
sapwood and they therefore require correction of measured Vh to account for the
damage caused by drilling and constant heating, as well as the influence on heat
transfer of the materials used to construct the temperature sensor (Green and
Clothier, 1988; Fernández et al., 2006; Steppe et al., 2010). For the CHPM the
correction can be made using empirical models derived by Swanson and Whitfield
(1981) and Green and Clothier (1988) for a variety of temperature sensors spacings
and materials used to construct the heater probes and temperature sensors. The
models are of the following form:
𝑉𝑐 = 𝑎 + 𝑏𝑉ℎ + 𝑐𝑉ℎ2 Equation 1.5
where Vc is corrected heat pulse velocity, and a, b and c are coefficients calculated
by the models for various wound widths. The models used to estimate these
Bark + Cambium Centre of stem Heartwood
Sapwood
Sap flow
Heater Probe
d
Temperature
Sensors
© University of Pretoria
18
coefficients as derived by Swanson and Whitfield (1981) and/or Green and Clothier
(1988) cannot be solved at zero sap flow and; that only compounds the problem of
poor resolution at low sap velocities using this method (Closs, 1958; Marshall, 1958;
Barrett et al., 1995).
Burgess et al. (2001) developed models for wound correction of the following form:
𝑉𝑐 = 𝑏𝑉ℎ + 𝑐𝑉ℎ2 + 𝑑𝑉ℎ
3 Equation 1.6
where b, c and d are coefficients calculated by the models for various wound widths.
The models developed by Burgess et al. (2001) for wounding correction to be used
in conjunction with the HRM are theoretically capable of estimating zero sap flow.
1.1.2.5 Calibration of the heat pulse techniques
Calibration of the heat pulse techniques is necessary for species with wood that are
not thermally homogeneous (Smith and Allen, 1996). Thermally homogeneous
sapwood should have both average xylem vessel lumen diameter and average
distance between the xylem vessels of less than 0.4 mm (Swanson, 1983). Sap flow
methods have been calibrated against other established methods to determine
adjustment coefficients (Green and Clothier, 1988; Steppe et al., 2010). These
established methods include weighing lysimeters (Caspari et al., 1993; Nortes et al.,
2008), whole tree gas exchange systems (Dragoni et al., 2005; Pernice et al., 2009),
cut tree experiments (Green and Clothier, 1988; Hatton et al., 1995; Fernández et
al., 2001), stem perfusion experiments (Green and Clothier, 1988; Fernández et al.
2006; Steppe et al., 2010) and micrometeorological measurements (Kostner et al.,
1992; Williams et al., 2004; Conceição and Ferreira, 2008; Poblete-Echeverría et al.,
2012).
Of all the parameters required to convert Vh to transpiration the wound width (i.e. the
damage caused by drilling and constant pulse heating) is difficult to estimate using
experimental procedures. A number of calibrations have shown that once all other
parameters have been experimentally determined, an apparent wound width can be
used to convert Vh measurements to transpiration in sapwood that is not thermally
homogeneous (Green and Clothier 1988; Zreik et al., 2003; Fernández et al., 2006).
Fernandez et al. (2006) termed this apparent wound width that could be used to
© University of Pretoria
19
convert Vh’s to transpiration so that it matched transpiration measure with an
independent method a “virtual wound width”.
There seem to be contradicting views on the development of the wound width when
heat pulse techniques are used for long-term transpiration measurements. Early
research findings reported that wounding effects increase with time and
recommended that Vh probes should be moved regularly, either to a different part of
the same stem or to a new stem (Swanson and Whitfield, 1981; Marshall et al.,
1981; Smith and Allen, 1996; Zreik et al., 2003). To the contrary other authors
showed that only limited changes occurred in xylem properties during a long period
of Vh measurements in olive trees (Fernández et al., 2001; Giorio and Giorio, 2003;
Pernice et al., 2009). Dragoni et al. (2005) also suggested that a single calibration
was sufficient for long-term transpiration measurements in an apple orchard.
1.1.2.6 Scaling from sample trees to orchard level
In orchard situations sap flow measurements are normally conducted on a few
representative trees from which whole stand transpiration is then extrapolated. The
scaling of transpiration from sample trees to stand level can be based on leaf area
(Jara et al., 1998), variability of plant stem diameter (Daamen et al., 1999; Bethenod
et al., 2000) or plant density (Dugas and Mayeux Jr, 1991; Dragoni et al., 2005).
Irrigated fruit tree orchards constitute clonal trees that are genetically identical and
are normally trained similarly. The trees are mostly the same size, receive almost
equal amounts of radiant energy and receive equal amounts of water from a uniform
irrigation system. Little variation in transpiration of individual trees is expected (Smith
and Allen, 1996) such that stand transpiration can be estimated from the sample
trees using planting density (e.g. Dragoni et al., 2005). Use of allometric
relationships between sap flow of sample trees growth parameters (i.e. stem
diameter, stem basal area, sapwood area or leaf area) (e.g. Allen and Grime, 1995;
Soegaard and Boegh, 1995; Vertessy et al., 1995) may not hold true since the
canopies are manipulated. However, in studies of Cohen et al. (2008) and
Conceição and Ferreira (2008) transpiration of sample trees was successfully scaled
to whole stand transpiration as a weighted mean based on stem diameter of the
sample trees in the orchards.
© University of Pretoria
20
1.1.3 Micrometeorological estimates of evapotranspiration
The EC is one micrometeorological method that is normally deployed into the
equilibrium boundary layer of a cropped area to estimate ETc (Allen et al., 2011).
Micrometeorological methods require large flat fields for the establishment of fetch
and minimum equipment height equipment to produce representative data (Paw et
al., 1995). Generally, instrumentation used for micrometeorological methods is
relatively fragile and expensive, such that constant technical attention is required;
and hence they are usually deployed into the field for short periods of time.
1.1.3.1 Eddy covariance method
Eddy covariance systems have been widely used to estimate ETc in orchards e.g.
olive (Testi et al., 2004; Villalobos et al., 2000; Pernice et al., 2009), apple (Jones et
al., 1988), pecan (Samani et al., 2011), mango (Teixeira and Bastiaanssen, 2012)
and citrus (Rana et al., 2005; Villalobos et al., 2009; Consoli et al., 2006) orchards.
The theory behind these measurements is based on the principle that turbulence in
the atmospheric boundary layer dominates mixing and the diffusion of sensible and
latent heat to and from the underlying surface (Swinbank, 1951). This requires high
speed measurement of sonic temperature, wind velocity, and water vapour pressure
or specific humidity, usually at frequencies of 5–20 Hz, using quick response
sensors. ETc is measured based on the high frequency of the turbulent ‘eddy flux’
motions in the surface boundary using the following statistical relationship of
Swinbank (1951):
𝐸𝑇𝑐 = 𝜌𝑎𝑤′𝑞′ =0.622
𝑝𝑤′𝑒′ Equation 1. 7
where ρa is density of moist air, p is atmospheric pressure, q’ is the instantaneous
deviation of specific humidity from mean specific humidity, e’ is the instantaneous
deviation of vapour pressure from mean vapour pressure, and w’ is the
instantaneous deviation of vertical wind velocity from mean vertical wind velocity.
The over-bar signifies a time average over a specified interval of time and the prime
indicates a departure from the mean. Sensors must measure vertical wind speed,
sonic temperature and humidity with sufficient frequency response to record the
© University of Pretoria
21
most rapid fluctuations important to the diffusion process. Wind speed and humidity
sensors should be installed close to each other but separated sufficiently to avoid
interference because if the separation is too large, an underestimate of the flux may
result (Lee and Black, 1994). Many corrections are applied to arrive at the actual flux
measurements.
The wide use of the EC method has been attributed to a number of reasons which
include: relatively easy to setup, reduced cost of sensors and the ability to co-
measure energy balance components (as well as carbon dioxide fluxes depending
on equipment configuration) (Allen et al., 2011). ETc measurements based on the EC
method can be done using a direct measurement or calculated as a residual in the
shortened energy balance.
1.1.3.1.1 Direct measurement
Instrumentation for direct measurement of water fluxes using the EC method
involves the use of sonic anemometers and rapid-response open-path infra-red gas
analysers (Tanner, 1988; Tanner et al. 1993; Wilson et al., 2002; Baldocchi, 2003;
Shaw and Snyder, 2003; Meyers and Baldocchi, 2005). Using these instruments
fluxes of latent (λETc) and sensible heat (H) are measured with a high level of
precision and with a high degree of spatial and temporal resolution averaging 15-30
minute periods. Typically, a frequency of the order of 5–10 Hz is used, but the
response-time requirement depends on wind speed, atmospheric stability and the
height of the instrumentation above the surface.
1.1.3.1.2 Energy balance method
ETc can be calculated as a residual of the energy balance once the measurement of
other components of the energy balance has been determined. The shortened
energy balance equation can be written as follows:
𝜆𝐸𝑇𝑐 = 𝑅𝑛 − 𝐺 − 𝐻 Equation 1.8
where λETc is the latent heat flux, Rn is the net radiation, G is the heat flux transfer
into the soil, H is the sensible heat flux density. All terms have units of W m-2. Rn is
© University of Pretoria
22
measured using net radiometers and G is measured using soil heat flux plates. The
sensible heat flux is measured using the EC method based on the following
statistical relationship of Swinbank (1951):
𝐻 = 𝜌𝑎𝐶𝑝𝑤′𝑇′ Equation 1.9
where T’ is the instantaneous deviation of air temperature from mean temperature
and Cp is specific heat of moist air. The use of the shortened energy balance has an
advantage of eliminating the requirement for the quick response hygrometer, which
can be expensive and can create high frequency fallout caused by physical
separation of the hygrometer from the sonic anemometer. The disadvantage of
estimating ETc as a residual of the shortened energy balance is the need to measure
Rn and G accurately, which can be problematic under conditions with sparse or
heterogeneous vegetation such as fruit tree orchards (Burba and Anderson, 2008). It
is also assumed that other terms of the full energy balance such as canopy stored
heat flux, advection flux e.t.c are negligible.
There are several corrections that must be applied to measurements because
practical instrumentation cannot fully meet the requirements of the underlying
micrometeorological theory (Foken et al., 2012). Typically, measurements are made
in a finite sampling volume rather than at a single point, and the maximum frequency
response of the sensors is less than the highest frequencies of the turbulent eddies
responsible for the heat and mass transport. Both of these cause a loss of the high-
frequency component of the covariances used to calculate fluxes. Errors also arise in
calculating fluxes of trace gas quantities using open-path analyzers because of
spurious density fluctuations arising from the fluxes of heat and water vapour. The
several corrections required on the sampled high frequency data for the EC method
to ensure that quality data are done using software such as EdiRe (Mauder et al.,
2008).
When the shortened energy balance is used it also must be checked for energy
balance closure error (Allen et al., 2011).This error occurs when the sum of
measured λETc + H does not equal measured Rn – G and can be in the range of 10
to 30 % for field experiments (Aubinet et al. 2000; Wilson et al. 2002). Foken et al.
(2008) listed the following as the possible causes for lack of closure (i) different
© University of Pretoria
23
reference levels and different sampling scales of the measuring methods for net
radiation, turbulent fluxes, and soil heat flux (ii) heterogeneous landscape in the
vicinity of a flux-measurement site that is capable of generating eddies at larger time
scales, that can be detected by the EC method (iii) advection and fluxes associated
with longer wavelengths, and (iv) measurement errors. Energy balance closure is
usually achieved by scaling λETc and H in the same proportion (Twine et al., 2000)
or using multiple linear regression equations (Allen et al., 2008).
1.2 Modelling citrus water use
1.2.1 Introduction
Variability in measured ETc of citrus orchards means that these measurements
cannot be directly applied to all areas in which the crop is cultivated and under all
management practices. In addition measurements of ETc are expensive, time
consuming and requires specialised expertise which is not always available. Models
play a pivotal role in extrapolating these measurements to regions in which they
were not conducted. ETc is modelled on a physical basis using weather data and
algorithms that describe the crop’s surface and aerodynamic resistances (Allen et
al., 1998). According to Allen et al. (1998) the Penman-Monteith equation has been
used quite extensively. The form of the equation is as follows:
𝜆𝐸𝑇𝑐 =∆(𝑅𝑛−𝐺)+𝜌𝐶𝑝
𝑉𝑃𝐷
𝑟𝑎
∆+𝛾(1+𝑟𝑠𝑟𝑎
) Equation 1.10
where λETc is the evapotranspiration rate per unit leaf area (W m-2), Rn is the net
radiation flux density absorbed by the leaf (W m-2), ρ and Cp, are the density (kg m-3)
and specific heat capacity (J kg-1 K-1) of air at constant pressure, λ is the latent heat
of vaporisation of water (J g-1), γ is the psychrometric constant (kPa oC), Δ is the
slope of the water vapour pressure curve (kPa oC) at ambient air temperature Ta
(oC), VPD is the saturation vapour pressure deficit of the air (kPa), ra is the
aerodynamic resistance (s m-l), and rs is the bulk surface resistance (s m-1).
The rs and ra terms used to simplify the exchange process in a vegetation layer in
Equation 1.10, are crop specific and cannot be determined easily (Figure 1.2). The rs
© University of Pretoria
24
term is used to describe the resistance of water flow through the soil-plant-
atmosphere continuum. The rs term combines resistances through the plant system
and ra describes the resistance from the surface of the vegetation upward and
involves friction from air flowing over vegetation surfaces (Figure 1.2). In order to
obviate the need to estimate the resistances parameters for each crop at different
stages of growth, the crop coefficient approach was introduced.
Figure 1.2. Simplified representation of the (bulk) surface and aerodynamic
resistances for water flow through the soil-plant-atmosphere continuum
(Allen et al., 1998).
1.2.2 Crop coefficient approach
The crop coefficient approach that was first published by Doorenbos and Pruitt
(1977) and later refined by Allen et al. (1998) has been used quite extensively in
modelling ETc, such that it is considered a standard for estimating ETc. ETc under
standard conditions is calculated as a product of experimentally determined crop
coefficients (Kc) and reference evapotranspiration (ETo). Standard conditions imply
that the crops are disease free, well fertilised, growing in large fields, under optimum
soil water conditions, and achieving full production under the given climatic
conditions. The method has been accepted in the field of irrigation water
management because it is relatively simple to use. The technical advantage of the
crop coefficient approach is that it makes it possible to consider the independent
© University of Pretoria
25
contributions of evaporation and transpiration (Allen et al., 1998). This is done by
splitting the Kc factor into two separate coefficients namely soil evaporation
coefficient (Ke) and a basal crop coefficient (Kcb).
1.2.2.2 Single crop coefficient
In the single crop coefficient approach, the individual contributions of transpiration
and evaporation are combined to determine ETc as follows:
𝐸𝑇𝑐 = 𝐾𝑐 × 𝐸𝑇𝑜 Equation 1.11
The Kc coefficient averages evaporation and transpiration and should be used to
compute ETc for weekly or longer time periods for irrigation water management,
where the averaged effects of soil wetting are acceptable and relevant e.g. in furrow
and sprinkler systems (Allen et al., 2007).
1.2.2.3 Dual crop coefficient
The dual crop coefficient allows for the separation of evaporation and transpiration
such that Equation 1.12 can be written as:
𝐸𝑇𝑐 = (𝐾𝑐𝑏 + 𝐾𝑒) × 𝐸𝑇𝑜 Equation 1.12
Allen et al. (1998) defined the Kcb as the ratio of ETc to ETo when the soil surface
layer is dry, but where the average soil water content of the root zone is adequate to
sustain full plant transpiration. This is best suited for irrigation water management in
micro-irrigation systems where high frequency irrigation is practised and large
variations in day to day soil wetness are experienced.
1.2.2.4 Reference evapotranspiration
Reference evapotranspiration may be expressed based on two reference crops,
clipped, cool-season grass or tall alfalfa, whose common symbols are ETo and ETr,
respectively. One, a 0.12-m tall cool-season clipped grass, follows the definition by
FAO-56 (Allen et al., 1998, 2006) and uses the PM equation with specific
parameterizations that now differ from the original 1998 FAO-56 definition (Figure
1.3) only in the value for bulk surface resistance prescribed for hourly or shorter
© University of Pretoria
26
time-step calculations, where rs was reduced (Allen et al., 2006) from the daily value
of 70 s m−1 to a new value of 50 s m−1 for hourly or shorter periods during daytime.
For ETr the standardized reference surface is the ‘tall’ reference, defined to be very
similar to a 0.5 m tall crop of dense, full-cover alfalfa (lucerne) having surface
resistance of 30 s m−1 for hourly or shorter calculation time-steps during daytime and
45 s m−1 for daily calculation time-steps (Pereira et al., 2015). Crop coefficients (Kc)
may be expressed in relation to clipped, cool-season grass as more often used
(Allen et al., 1998; 2007) or to alfalfa; for which the symbol Kcr is adopted (ASCE-
EWRI, 2005; Allen et al., 2007).
Figure 1. 3. Characteristics of the hypothetical short grass reference crop used for
calculation of ETo in the FAO Penman-Monteith equation (Allen et al.,
1998).
The Penman-Monteith equation standardised for the determination of ETo takes the
following form (Allen et al., 2006; Allen et al., 2007; Pereira et al., 2015):
𝐸𝑇𝑜 =0.408∆(𝑅𝑛−𝐺)+𝛾
𝐶𝑛𝑇𝑎+273
𝑢2𝑉𝑃𝐷
∆+𝛾(1+𝐶𝑑0.34𝑢2) Equation 1.13
where ETo is reference evapotranspiration (mm day-1), Rn is net irradiance at the
crop surface (MJ m-2 day-1), G is soil heat flux density (MJ m-2 day-1), Ta is mean
© University of Pretoria
27
daily air temperature at 2 m height (°C), u2 is wind speed at 2 m height (m s-1), VPD
is saturation vapour pressure deficit (kPa), Δ is slope of the vapour pressure curve
(kPa °C-1), γ is the psychrometric constant (kPa °C-1), Cn is a numerator constant
that changes with reference type and calculation time step and Cd is a denominator
constant that changes with reference type and calculation time-step. The values for
Cn and Cd are presented in Table 1.2. The values for Cn consider the time-step and
aerodynamic roughness of the surface (i.e., reference type). The constant in the
denominator, Cd, considers the time-step, bulk surface resistance, and aerodynamic
roughness of the surface.
Table 1.2. Values for Cn and Cd in equation 1.13 for use when determining
evapotranspiration for short (ETo) and tall (ETr) reference crop (Allen et
al., 2007)
Calculation
time-step
ETo ETr Units for
ETo, ETr
Units for
Rn, G
Cn Cd Cn Cd
Daily 900 0.34 1600 0.38 mm d-1 MJ m-2 d-1
Hourly
(daytime)
37 0.24 66 0.25 mm h-1 MJ m-2 h-1
Hourly
(nighttime)
37 0.96 66 1.7 mm h-1 MJ m-2 h-1
1.2.2.5 Citrus crop coefficients
The standard crop coefficients for citrus published by Allen et al. (1998), which are
largely based on the work of Doorenbos and Pruitt (1977), used for estimating ETc
are presented in Table 1.3. In developing the crop coefficients (Kc and Kcb), the
growing season is divided into four distinct growth stages (i.e. initial, crop
development, mid-season and late season stages) that are related to changes in
water use due to crop development. For a generic citrus crop Allen et al. (1998)
stipulated the lengths of the initial, development, mid-season and late season growth
stages as 60, 90, 120 and 95 days, respectively. However, only the crop coefficients
© University of Pretoria
28
during the initial, mid-season and late season presented in the table are required to
construct the crop coefficient curve.
Table 1.3. Citrus crop coefficient (Kc and Kcb) values for standard climate of RHmin =
45 % and u2 = 2 m s-1 as published for use with FAO Penman-Monteith
ETo (Allen et al. 1998).
Orchard and
ground conditions
Kc ini Kc mid Kc end Kcb ini Kcb mid Kcb end
No ground cover
70 % canopy 0.7 0.65 0.7 0.65 0.6 0.65
50 % canopy 0.65 0.6 0.65 0.6 0.55 0.6
20 % canopy 0.5 0.45 0.55 0.45 0.4 0.5
Active ground cover
70 % canopy 0.75 0.7 0.75 0.75 0.7 0.75
50 % canopy 0.8 0.8 0.8 0.75 0.75 0.75
20 % canopy 0.85 0.85 0.85 0.8 0.8 0.85
NB: subscripts ini, mid and end are incorporated to mean initial, mid-season and end stages of crop growth.
Citrus crop coefficients presented in Table 1.3 either decrease during the mid-
season or are constant throughout the season. This is in contradiction to the general
crop coefficient curve for annual crops where the highest values occur during the
mid-season. The occurrence of the highest crop coefficients during this period in
most crops is probably due to the fact that leaf area of these crops is highest at this
point, which translates into maximum water use at this time. The decrease of citrus
crop coefficient values during the mid-season may be interpreted to mean that as the
atmospheric demand increases, transpiration does not increase at the same rate
(Green and Moreshet, 1984; García Petillo and Castel, 2007; Villalobos et al., 2009).
This behaviour has been linked to strong regulation mechanisms within the citrus
tree hydraulic path (Kriedemann and Barrs, 1981).
The adoption and subsequent use of the FAO-56 crop coefficients in citrus is
regardless of the fact that a wide range of crop coefficients has been reported for the
crop in different growing regions (Table 1.4). Three significant deductions can be
© University of Pretoria
29
made from the data presented in Table 1.4. Firstly, the crop coefficients of Allen et
al. (1998) are all less than one, as opposed to some studies in Table 1.4. Secondly,
the values reported by some of the authors are much lower than those reported by
Allen et al. (1998). Thirdly, some of the crop coefficients referred to in Table 1.4 are
lower in summer (period normally coinciding with the mid-season, which is
associated with active tree growth and fruit production) than in winter which agrees
with the crop coefficients of Allen et al. (1998). According to Goldhamer et al. (2012)
this is generally attributed to the citrus stomata being sensitive to evaporative
demand (the air vapour pressure deficit, VPD), closing under dry, hot, windy
conditions and opening under the opposite conditions. Thus, the Kc in the mild
Mediterranean and coastal climates should be higher than those of more arid, inland
valleys. In addition, in Mediterranean environments, high Kc values in winter reflect
high soil evaporation rates from frequent rainfall during that part of the season.
The factors that affect the water use of citrus trees that were mentioned above will
have an inherent effect on the crop coefficients. These factors include differences in
variety, rootstock, tree spacing, canopy height, ground cover, tillage, leaf area index,
method of estimating reference evapotranspiration, microclimate, irrigation method
and frequency and method of measuring crop evapotranspiration (Snyder and
O’Connell 2007; Naor et al. 2008). Another source of variation that is particular to Kcb
values, that are based on transpiration determined using sap flow methods (i.e.
Rana et al. 2005; Villalobos et al., 2009; Marin and Angelocci, 2011; Villalobos et al.,
2013), is that the Kcb values of Allen et al. (1998) represent the transpiration
component of ETc and a small evaporation component from soil that is visibly dry at
the surface. Although the evaporation component is small, it is not detected by sap
flow methods. To distinguish the two forms, Villalobos et al. (2013) suggested that
sap flow-based crop coefficients should be termed transpiration coefficients (Kt).
© University of Pretoria
30
Table 1.4. Seasonal crop coefficients determined in citrus orchards across the growing regions of the world.
Reference Region ETo Kc
Summera Autumn Winter Spring Summer Autumn Winter Spring
Villalobos et al. (2013) Spain 0.34 0.5 0.27 0.3
Marin and Angelocci (2011) Brazil 4.4 2.7 0.7 0.26
Villalobos et al. (2009) Spain 4.99 5.88 0.44 0.43
Fares et al. (2008) USA 4.40 2.24 0.86 0.69
Snyder and O'Connell (2007) USA 6.27 3.03 1.10 4.13 1.00 1.07 1.15 1.13
Alves et al. (2007) Brazil 4.33 3.49 2.90 4.33 1.05 0.93 0.53 0.69
García Petillo and Castel (2007) Uruguay 4.99 1.77 1.45 4.06 0.64 0.77 0.84 0.83
Romero et al. (2006) Spain 0.7 0.7 0.9 0.7
Rana et al. (2005) Italy 1.04 0.77 0.81 1.16
Yang et al. (2003) Japan 4.40 0.60 0.90 0.60
Castel (1997) Spain 4.69 2.75 1.66 3.68 0.32 0.49 0.36 0.24
Castel et al. (1987) Spain 4.93 2.57 1.53 3.43 0.70 0.77 0.65 0.61
Rogers et al. (1983) USA 4.1 3.13 2.37 4.37 1.04 1.01 0.93 0.81
Hoffman et al. (1982) USA 7.7 4.1 2.27 5.27 0.85 0.83 0.77 0.80
Green and Moreshet (1979) South Africa 6.86 3.15 2.62 4.59 0.80 1.28 0.79 0.62
van Bavel et al. (1967) USA 7.91 3.77 2.25 6.30 0.62 0.72 0.48 0.48
aSeasons were determined according to the equinoxes and solstices, with each season comprising 3 months. Measured data
available for the period in the specific seasons was used for each cited work.
© University of Pretoria
31
Most of the works reported on crop coefficients in citrus in Table 1.4 (i.e. Castel,
1997; Rana et al., 2007; Marin and Angelocci, 2011; Villalobos et al., 2009; Consoli
and Papa, 2013; Villalobos et al., 2013; Marin et al., 2016), and other fruit tree
orchards (Ayars et al., 2003; Girona et al. 2003; Testi et al. 2004; Girona et al., 2005;
Johnson et al., 2005; Paco et al., 2006), have used ETo. This is despite of the fact
that these tall tree crops with a large roughness that can probably be best described
by ETr rather than ETo. The use of the ETo as the principal reference method has
been attributed to its early adoption in FAO24 and the simplicity in cultivating this
reference (clipped, cool season grass) across a wide range of locations and climates
(Pereira et al., 2015). Generally, ETr is about 1.1 to 1.4 times that of ETo due to the
increased roughness and leaf area of alfalfa.
In an attempt to improve on the accuracy of crop coefficients when transferred
amongst different orchards Allen and Pereira (2009) consolidated and formalised the
FAO-56 procedure for estimating the crop coefficients as a function of fraction of
ground cover and crop height.
1.2.3 Estimation of transpiration crop coefficients from fractional ground cover
and crop height
Allen et al. (1998) suggested a procedure for estimating the crop coefficients as a
function of fraction of ground cover and crop height using a plant density coefficient.
The advantage of this method is that it is relatively simple and is based on physical
measurements that can easily be carried out in the field. Allen and Pereira (2009)
further consolidated the concept to enable the estimation of site-specific crop
coefficients based on the fraction of ground covered or shaded by the vegetation, the
height of the vegetation and the degree of stomatal regulation under wet soil
conditions.
Allen and Pereira (2009) conceived the idea of an upper, energy-constrained limit on
Kc that can be adjusted for vegetation without full ground cover, such as orchards.
They defined this upper limit, termed Kc max, as the maximum value for Kc following
rain or irrigation. The value for Kc max is governed by the amount of energy available
for evaporation of water, which is largely encapsulated in ETo. The Kc max varies with
© University of Pretoria
32
general climate, ranging from about 1.05 to 1.30 (Allen et al., 1998; Allen et al.,
2005) as follows:
𝐾𝑐 𝑚𝑎𝑥 = 𝑚𝑎𝑥 ({1.2 + [0.04(𝑢2 − 2) − 0.004(𝑅𝐻𝑚𝑖𝑛 − 45)] (ℎ
3)
0.3
} , {𝐾𝑐𝑏 + 0.05})
Equation 1.14
where RHmin is average daily minimum relative humidity during the growth state or
period and h is the mean plant height (m) during the period of calculation (initial,
development, midseason, or late-season).
Transpiration increases with plant density or leaf area, so does the Kc value (Allen
and Pereira, 2009) until the optimum is reached at a value of maximum canopy
cover, which is usually defined as a leaf area index (LAI) equal to 3. In situations
where full canopy cover is not achieved the Kcb, which correlates with the amount of
vegetation because it represents mostly transpiration, can be expressed in terms of
a density coefficient, Kd, where:
𝐾𝑐𝑏 = 𝐾𝑐 𝑚𝑖𝑛 + 𝐾𝑑(𝐾𝑐𝑏 𝑓𝑢𝑙𝑙 − 𝐾𝑐 𝑚𝑖𝑛) Equation 1.15
where Kcb is the approximation for the basal crop coefficient for conditions
represented by the density coefficient, Kd, Kcb full is the estimated basal Kc during
peak plant growth for conditions having nearly full ground cover (or LAI = 3), and Kc
min is the minimum basal Kc for bare soil (Kcb min is approximately equal to 0.15 under
typical agricultural conditions).
1.2.3.1 Basal crop coefficient during peak plant growth (Kcb full)
Kcb full for large stand size (greater than about 500 m2) can be approximated as a
function of mean plant height and adjusted for climate following Allen et al. (1998)
as:
𝐾cb full = 𝐹r (min(1.0 + 0.1ℎ, 1.20) + [0.04(𝑢2 − 2) − 0.004(𝑅𝐻min − 45)] (ℎ
3)
0.3
)
Equation 1.16
where Fr [0-1] is a relative adjustment factor for stomatal control, h is mean monthly
plant height (m), Parameter Fr applies a downward adjustment (Fr ≤ 1.0) if the
vegetation exhibits more stomatal control on transpiration than is typical of most
annual agricultural crops, a situation that is typical of citrus (Kriedemann and Barrs,
1981). Allen and Pereira (2009) suggested the following calculation for Fr for full
© University of Pretoria
33
cover vegetation, based on the FAO Penman-Monteith equation and assuming full
cover conditions:
𝐹𝑟 ≈∆+𝛾(1+0.34𝑢2)
∆+𝛾(1+0.34𝑢2𝑟𝑙
100) Equation 1.17
where rl is mean leaf resistance for the vegetation in question (s m-1); ∆ is the slope
of the saturation vapour pressure versus air temperature curve (kPa °C-1) and γ is
the psychrometric constant (kPa °C-1). For most annual agricultural crops the value
of rl is 100 s m-1, which sets Fr to 1. Allen and Pereira (2009) suggest a value of 420
s m-1 for the initial and midseason periods and 150 s m-1 at the end of the season for
citrus.
1.2.3.2 Estimation of plant density coefficient (Kd)
According to Allen and Pereira (2009), Kd describes the increase in Kc with increase
in amount of vegetation. The shape of the Kd curve is curvilinear with LAI or fraction
of ground cover because of effects of micro-advection of convective and radiative
energy from exposed soil and height of vegetation. The density coefficient Kd can be
estimated as a function of measured or estimated LAI or as a function of fraction of
ground covered by vegetation.
1.2.3.2.1 Estimation of Kd from LAI
Where LAI can be measured or approximated, Kd can be approximated under
normal conditions using an exponential function by suggested by Allen et al. (1998)
as follows:
𝐾𝑑 = 1 − 𝑒0.7𝐿𝐴𝐼 Equation 1.18
where LAI is defined as the area of leaves per area of ground surface averaged over
a large area with units of m2 m-2. Only one side of green healthy leaves that are
active in vapour transfer is counted.
1.2.3.2.2 Estimation of Kd from ground covered by vegetation
Kd can also be estimated from measurement of ground covered by the vegetation
using the following formula suggested by Allen et al. (1998):
𝐾𝑑 = min (1, 𝑀𝐿𝑓c eff, 𝑓c eff
(1
1+ℎ)) Equation 1.19
© University of Pretoria
34
where ƒc eff is the effective fraction of ground covered or shaded by vegetation [0.01-
1] near solar noon, ML is a multiplier on ƒc eff and is an attempt to simulate hydraulic
resistances within the plant and is expected to range between 1.5-2.0, with a value
of 1.5 recommended for citrus (Allen and Pereira, 2009).
Of the two methods for determining Kd, Equation 1.19 is of practical importance due
to the ease involved in determining fc eff and h. For canopies such as trees fc eff can
be estimated according to Allen et al. (1998) as follows:
𝑓c eff =𝑓𝑐
sin 𝛽≤ 1 Equation 1.20
where β is the mean angle of the sun above the horizon during the period of
maximum ETc and ƒc is the fraction of surface covered by vegetation as observed
from directly overhead. β at solar noon can be calculated as:
𝛽 = arcsin[sin(𝜑) sin(𝛿) + cos(𝜑) cos(𝛿)] Equation 1.21
where φ is latitude and δ is solar declination in radians.
For ready application of the methodology, Allen and Pereira (2009) recommended a
set of generic parameters to be used in conjunction with Equations 1.16, 1.17 and
1.19 for different crops including citrus. Generic parameters for citrus orchards are
presented in Table 1.5. Typical crop coefficients developed from effective ground
cover that have been published for mature citrus orchards are presented in Table
1.6. The rl values in Table 1.5 were estimated by inverting Equation 1.17 after
solving for Fr by inverting Equation 1.16 using known Kcb full values. This procedure
aimed at producing values for Kc and Kcb that agree with literature values, including
those from FAO-56. It is emphasised that values for rl determined by this inversion
are only for reuse in Equation 1.17. Allen and Pereira (2009) advised that rl
computed in this way must be evaluated with caution, and recommend the need to
conduct research to improve the estimation of this parameter.
Nearly all rl values obtained for orchards and vine crops following this approach
exceeded the average value of 100 s m-1 normally associated with annual
agricultural vegetation. This was interpreted by Allen and Pereira (2009) as an
indication of strong degree of stomatal control exhibited by different fruit tree species
under typical growing conditions. In a comparative study Meyer and Green (1981)
© University of Pretoria
35
observed that leaf resistances of citrus trees were much lower than that observed in
soybean and wheat grown under similar environmental conditions. Average leaf
resistances higher than what was suggested by Allen and Pereira (2009) have also
been reported in citrus species e.g. 500 to 2000 s m-1 in Shamouti orange trees
(Cohen and Cohen, 1983), 200 to 8280 s m-1 in Navel orange trees (Dzikiti et al.,
2007) and 295 and 830 s m-1 in oranges (Pérez-Pérez et al., 2008).
Table 1.5. Generic parameters for a citrus crop for estimating crop coefficients.
Orchard ground
conditions
Parameters at initial
and midseason
periods
Parameters at end
of season
ML h (m) fc rl Fr fc rl Fr
No ground
cover
1.5 2 0.25 420 0.71 0.20 150 0.94
Ground cover 1.5 2.5 0.45 420 0.71 0.40 275 0.82
Table 1.6. Citrus values of Kcini, Kc mid, Kc end, Kcbini, Kcb mid and Kcb end for standard
climate of RHmin = 45 % and u2 = 2 m s-1 as expanded from FAO-56
(Allen et al. 1998) for a range of values for effective ground covered by
the tree canopy (fc eff) during the midseason and using the procedure of
Allen and Pereira (2009) with recommended parameter values.
Tree density and
ground conditions
Kcini Kc mid Kc end Kcbini Kcb mid Kcb end
No ground cover
High (fc eff = 0.70) 0.95 0.90 90 0.85 0.85 0.85
Medium (fc eff = 0.50) 0.80 0.75 0.75 0.70 0.70 0.70
Low (fc eff = 0.25) 0.55 0.45 0.45 0.40 0.40 0.40
Active ground cover
High (fc eff = 0.70) 1.00 0.95 0.95 0.90 0.90 0.90
Medium (fc eff = 0.50) 0.95 0.95 0.95 0.85 0.90 0.90
Low (fc eff = 0.25) 0.90 0.90 0.90 0.80 0.85 0.85
© University of Pretoria
36
Citrus trees exhibit high resistances to water movement (van Bavel et al., 1967;
Cohen and Cohen 1983; Syvertsen and Graham 1985; Rodríguez-Gamir et al. 2010)
in the soil-plant-atmosphere continuum that can be detected at leaf level. The
resistance is an essential parameter for understanding the mechanisms of plant
transpiration in citrus orchards. Ideally it would be good to conduct extensive
measurements of leaf resistances using porometry. However, such measurements
are affected by a number of shortcomings which include: (i) sampling of leaf
resistances in the field are not only labour intensive but are further complicated by
the fact that rl varies with leaf age and climatic conditions (van Bavel et al., 1967); (ii)
automation of such measurements cannot be made because the sensor creates an
artificial environment, which may lead to changes in the physiological properties of
the leaf; and (iii) manual measurements performed with hand held porometers do not
allow simultaneous sampling of the leaves exposed to varying levels of solar
radiation by one user (Jarvis, 1976).
When measurements of water use are available, the bulk resistance of a vegetated
surface can be obtained using the top–down approach by inversion of Equation 1.10
(Baldocchi et al., 1991). Transpiration measured using sap flow methods have been
used quite extensively to determine bulk resistances in this manner (Wullschleger et
al., 1998). Researchers have, however, reported these in terms of the bulk
conductance rather than resistances. This is because flux is directly proportional to
the conductance, such that errors in conductance, rather than the resistances are
normally distributed (Campbell and Norman, 2008). The general approach has been
to use such data sets to parameterise models that can predict canopy conductance
as a function of canopy size and environmental variables (e.g. Zhang et al., 1997;
Oguntunde et al., 2007; Villalobos et al., 2009; Granier and Loustau 1994; Green
and McNaughton 1997) for predicting transpiration using Equation 1.10.
1.2.4 Modelling canopy conductance
Stomatal conductance of illuminated leaves has been shown to be dependent upon
current levels and history of the immediate environmental variables, which includes
radiant flux density, ambient carbon dioxide concentration, leaf air vapour pressure
difference and leaf water status (Jarvis, 1976). A number of models have been
© University of Pretoria
37
formulated to describe the behaviour of stomatal conductance in relation to these
environmental variables (Damour et al., 2010). Conductance models can be largely
divided into multiplicative (Jarvis-type) and multiple linear regression models.
1.2.4.1 Jarvis-type model
Jarvis (1976) proposed the first multiplicative model involving radiant flux density
(SR), ambient carbon dioxide concentration (Ca), leaf-air water vapour pressure
difference (VPD), air temperature (Ta) and leaf water status (Ψl), which is of the
following form:
𝑔𝑠 = 𝑔𝑠 𝑚𝑎𝑥ƒ(𝑆𝑅)ƒ(𝐶𝑎)ƒ(𝑉𝑃𝐷)ƒ(𝑇𝑎)ƒ(𝛹𝑙)
Equation 1.22
where gs is the stomatal conductance predicted by the Jarvis model (m s-1), gs max is
the maximum stomatal conductance (m s-1). This complex set of factors modulating
stomatal all at once makes it almost impossible to quantify the effects of each
environmental effect in the field. Controlled laboratory studies have been used to
understand the stomatal responses to each of these factors in citrus and other crops.
Citrus stomata were shown to increase with photosynthetically active radiation until
saturation at around 500 μmol m-2 s-1 (Kriedemann, 1971). Stomata of most plants
response to the increase in atmospheric carbon dioxide concentration by reducing
the stomatal aperture. Although transpiration is reduced photosynthetic assimilation
increases such that water use efficiency is increased. Such responses will have
implications for gas exchange of all terrestrial plants under future elevated
concentrations of carbon dioxide. According to Levy and Kaufmann (1976) citrus
stomata over react to a step-change increase in VPD by opening and closing
rhythmically even when environmental conditions remain constant. Significant
responses of citrus stomata to temperature were observed in the range 20 to 35 oC
with the maximum occurring at 30 oC (Khairi and Hall, 1976). It was also shown that
values of leaf water potential at which appreciable citrus stomatal closure occur (-2
500 to -3 000 kPa) are much lower than typical values averaging -2 300 kPa
observed in irrigated trees (Syvertsen and Lloyd, 1994). As such leaf water potential
does not have a direct impact on stomatal behaviour of well-irrigated citrus orchards.
Based on these facts Syvertsen and Lloyd (1994) concluded that a Jarvis model with
© University of Pretoria
38
functions of SR, Ta and VPD should be sufficient to describe the stomatal
conductance of well-irrigated citrus trees.
By assuming the ‘big-leaf’ theory the Jarvis-type models have been extended to
predict canopy conductance (gc). The model, however, has been parameterised
using the minimum number of weather variables that captures the variations of
canopy conductance, with satisfactory accuracy in several fruit tree species (e.g.
Stewart, 1988; White et al. 1999; MacFarlane et al., 2004; Misson et al., 2004; Noe
and Giersch, 2004; Villalobos et al., 2000; Cohen and Cohen, 1983; Oguntunde et
al., 2007). Results of the application of the Jarvis-type model in citrus orchards are
not consistent. Cohen and Cohen (1983) reported that a Jarvis-type model
containing SR, VPD and Ta poorly (R2 = 0.11) described stomatal resistance of citrus
trees. Oguntunde et al. (2007) on the other hand showed that a similar model could
describe the variations of canopy conductance fairly well (R2 > 0.7). Elimination of
the temperature function did not affect the performance of the model in the case of
Oguntunde et al. (2007).
It is imperative that the aerodynamic conductance (ga) term should be determined in
order for gc to be calculated or predicted using Equation 1.10. However, it has been
argued that in the case of well-ventilated canopies, such as orchards, the effect of ga
on transpiration is far less critical than that imposed by gc (Tan et al., 1978;
McNaughton and Jarvis, 1983). The methods that have been used to determine ga
are reviewed in the next section.
1.2.4.3 Aerodynamic conductance
The transfer of water vapour at the surface of the crop canopy is sustained by
molecular diffusion through a thin skin of air, known as the boundary layer, in contact
with the surface (Monteith and Unsworth, 1990). The resistance to water vapour
movement in this layer is known as boundary layer or aerodynamic resistance and;
is difficult to quantify under field conditions, such that they are usually estimated
using empirical models (Pereira et al., 2006). One model that has been used
frequently for trees with leaves subjected to mutual interference is that proposed by
Landsberg and Powell (1973). The original model is:
© University of Pretoria
39
𝑟𝑎 = 58𝑝0.56 (𝑑
𝑢)
0.5
Equation 1.23
where p is a measure of foliage density to wind, estimated by the ratio between the
tree leaf area and the area of its silhouette projected on to a vertical plane, d is a
characteristic leaf dimension (m), sometimes taken as the square root of the mean
leaf area, and u is the average wind speed (m s-1) at the mid-canopy height.
Rana et al. (2005), working in a Clementine orchard, also developed a model for
estimating boundary layer resistance, which was later used by Oguntunde et al.
(2007) in a citrus orchard. This method relies on indirect estimations of zero plane
displacement height, roughness length governing the transfer of heat and vapour,
roughness length governing momentum transfer, von Karman’s constant and wind
speed measurements. The requirement for wind speed measurements at a certain
height above the canopy is the major drawback to this method. The equation has the
following form:
𝑟𝑎 =𝑙𝑛(
𝑧−𝑑
𝑧𝑜)𝑙𝑛(
𝑧−𝑑
ℎ𝑐−𝑑)
𝑘2𝑢𝑧 Equation 1.24
where k is the von Karman’s constant equal to 0.4, uz is the wind speed (m s-1) at the
z wind measurement height (m), d is the zero plane displacement estimated as d =
0.67hc, zo is the roughness length taken as 0.1hc and hc is the mean orchard height.
Equation 1.23 is usually used by most researchers because of of its simplicity.
1.3 Scope of the study
Long-term studies of citrus ETc are needed to update such information in the South
African industry. This study focussed on the long-term measurement of tree
transpiration in three citrus orchards using the HRM sap flow technique developed
by Burgess et al. (2001). Calibration of the sap flow technique was performed using
the eddy covariance method during the dry season (i.e. winter and autumn) when
evaporation from the orchard floor was considered negligible. Calibration of the HRM
was done by establishing a ‘virtual wound width’ used to convert heat pulse
velocities to transpiration. Crop coefficients have been used quite extensively in
estimating crop water use including citrus orchards. However, the variations among
the recently published crop coefficients have cast doubt on their transferability
among different growing regions. Allen and Pereira (2009) formalised the FAO-56
© University of Pretoria
40
methodology for estimating site-specific crop coefficients using canopy height, a crop
density factor and leaf resistance. Citrus trees exhibit high resistances to water
movement and quantifying their resistance is an important step towards developing
mechanistic models of transpiration in these species. Transpiration measurement at
stem level presents the opportunity to determine canopy conductance, that are
difficult to quantify under field conditions (Blad et al., 1982; Allen et al., 1998), using
a top-down approach. The possibility of using a Jarvis-type canopy conductance
model in conjunction with the Penman-Monteith equation to predict seasonal citrus
transpiration was investigated in three orchards.
© University of Pretoria
41
Chapter 2: Calibration of the heat ratio method (HRM) sap flow technique for
long-term transpiration measurement in citrus orchards
2.1 Introduction
Accurate and reliable methods of measuring water use of fruit tree orchards are
needed to gather information that is necessary to improve available irrigation water
management tools. Amongst the many methods that have been used to measure
tree transpiration, thermometric sap flow techniques have shown considerable
promise (Smith and Allen, 1996; Wullschleger et al., 1998). Some of the advantages
that have led to the popularisation of these methods include that they are easily
interfaced to data loggers for direct automated transpiration measurement; their
applicability is not inhibited by complex terrain and soil heterogeneity within fruit tree
orchards; they allow an estimation of transpiration which has a direct relationship
with productivity; and their relatively high accuracy of measuring orchard water use
as compared to other conventional methods e.g. eddy covariance (Rana et al., 2005)
and soil water balance approaches (Nicolas et al., 2005).
Thermometric techniques that have been employed for sap flow measurements in
woody species can be broadly categorized into heat pulse, heat balance and heat
dissipation methods (Smith and Allen, 1996). Heat pulse velocity (Vh) techniques, the
compensation heat pulse method (CHPM) in particular, have been most widely used,
mainly because they are relatively easy to use and require less power in the field
(Edwards et al., 1997; Conejero et al., 2007). However, CHPM is associated with
limitations, of which the major one is the inability to measure low sap flow rates,
which usually occur at night and under water stress conditions (Becker, 1998).
Burgess et al. (2001) developed the heat ratio method (HRM) as an improvement to
the CHPM in order to cater for low and reverse rates of sap flow in woody species.
The use of this relatively new method of measuring transpiration has not been
independently tested in citrus orchards.
Heat pulse techniques are invasive in nature because the probes are inserted into
the sapwood and require correction for the damage, commonly referred to as
wounding, caused by drilling and constant heating (Swanson, 1994). Conversion of
measured Vh’s to sap volumetric flow is performed using numerical models as a
© University of Pretoria
42
function of a number of wood parameters, which include sapwood water content,
sapwood density and wound width. The most sensitive and difficult parameter to
determine using experimental procedures is the actual wound width (Smith and
Allen, 1996; Fernández et al. 2006; Poblete-Echeverria et al., 2012). Some studies
have used predetermined corrections based on the theoretical calibrations of
Swanson and Whitfield (1981) (e.g., Nicolas et al., 2005, Villalobos et al., 2013) with
visually determined wounding widths (Green and Clothier, 1988; Dragoni et al.,
2005) or empirical calibration factors (Green and Clothier, 1988; Steppe et al., 2010)
to correct for wounding.
It has been demonstrated that for sapwood that is not thermally homogeneous a
“virtual wound width” is usually required to convert heat pulse velocities to volumetric
sap flow with appreciable accuracy (Green and Clothier, 1988; Zreik et al., 2003;
Fernández et al., 2006). Thermal homogeneity has been based on the work of
Swanson (1983) who categorised it on the basis of the sapwood xylem vessels
(particularly the diameter and density). However, there seem to contradicting reports
on the response of wood tissue surrounding the inserted probes during long-term
transpiration measurements. Smith and Allen (1996) suggested that wounding
effects increase with time and the Vh probes should be moved regularly, either to a
different part of the same stem or to a new stem. On the contrary, some studies
seem to suggest that only limited changes occur in xylem properties during long
periods of Vh measurements in some tree species e.g., olives (Fernández et al.,
2001; Giorio and Giorio, 2003; Pernice et al., 2009) and apples (Dragoni et al.,
2005).
Sap flow measurements for determining individual tree transpiration have been
calibrated against independent methods of measuring tree transpiration, which
include weighing lysimeters (Caspari et al., 1993; Nortes et al., 2008), whole tree gas
exchange systems (Dragoni et al., 2005; Pernice et al., 2009), cut tree experiments
(Green and Clothier, 1988; Hatton et al., 1995; Fernández et al., 2001), stem
perfusion experiments (Green and Clothier, 1988; Fernández et al., 2006) and
micrometeorological measurements (Jones et al., 1988; Poblete-Echeverría et al.,
2012). The eddy covariance method has been used in a number of experiments to
© University of Pretoria
43
calibrate sap flow measurements in orchard situations (e.g., Conceição and Ferreira,
2008; Poblete-Echeverría et al., 2012).
Crop evapotranspiration (ETc) measurements using micrometeorological
measurements conducted in orchards, during dry periods with minimal ground cover
when soil evaporation is negligible have shown that water loss from the orchard is
predominantly due to tree transpiration (Rogers et al., 1983; Pernice et al., 2009;
Marin and Angeloci, 2011). ETc measurements conducted during such periods can
be used to calibrate sap flow methods in the field. ETc measurements using
micrometeorological measurements are usually done for short periods of time,
largely because they require constant technical attention by well-trained personnel
and instrumentation is fragile and expensive (Allen et al., 2011). There is therefore a
need to establish the minimum number of ETc measurements required for calibration
of the HRM when it is used for long-term transpiration measurements.
The aim of this work was to calibrate the HRM, for measuring transpiration, against
ETc, measured using the eddy covariance method, in citrus tree orchards during dry
periods when evaporation is assumed negligible. It was hypothesized that citrus
sapwood matrix is not thermally homogeneous such that conversion of measured
heat pulse velocity to transpiration requires an empirical calibration factor which will
be constant for long-term measurements.
2.2 Materials and methods
2.2.1 Experimental Site and Climate
Measurements were conducted in a commercial orchard in the 2008/9 season at
Moosrivier Farm (Schoeman Boerdery Group) about 15 km north of the town of
Groblersdal. The farm is located in the summer rainfall area of Mpumalanga
Province of South Africa (25o 02’ 32.69” S and 29o 22’ 09.76” E) with an altitude of
approximately 900 m.a.s.l. The area receives an average annual rainfall of 535 mm
(Lynch and Schulze, 2006) and has an average minimum air temperature of 12.2°C
and maximum air temperature of 24.6°C (Schulze and Maharaj, 2007). The orchard
was planted in 1997 with ‘Delta’ Valencia orange trees (Citrus sinensis L. Osbeck)
grafted on ‘Swingle’ rootstock, at a spacing of 5.5 m x 2.75 m in a north–south
© University of Pretoria
44
orientation (0° N). Measured leaf area index (LAI) of the orchard was 2.85 m2 m-2.
Every alternate tree was “sandwich” pruned, meaning this tree was pruned narrower
every year to allow for the trees on either side to grow into the pruned space.
“Sandwich” pruned trees are removed once the trees on either side have reached an
adequate size as determined by the grower. Average tree height was 4.1 m. The
orchard was drip irrigated with one line per tree row and pressure compensating
emitters with a discharge of 2 L h-1 spaced 1 m apart. The soil type was sandy loam
with an average of 7-12 % clay in the first 1 m depth. The orchard block was circular
in shape with an average diameter of 595 m which allowed a fetch of 300 m from the
prevailing northerly winds for micrometeorological measurements conducted at the
centre of the circular field (Figure 2.1).
Figure 2.1. Map showing the ‘Valencia’ orange orchard highlighted by the yellow
boarder line. The position of the tower during micrometeorological
measurements is indicated by the Δ on the map.
2.2.2 Sap flow measurements
2.2.2.1 Measurement of heat pulse velocities
Sap flow measurements were conducted at stem level using the heat ratio method
as described by Burgess et al. (2001) on three sample trees in the orchard. Selection
N
© University of Pretoria
45
of the three sample trees that were instrumented was based on a tree stem
circumference survey at the start of the study. Stem circumferences of 40 trees were
measured at a height of 40 cm above the ground and ranged from 32.7 to 47.7 cm.
The results of the survey were then fitted to a normal distribution to establish suitable
circumference ranges to be used and select representative trees to be instrumented.
Table 2.1 shows the ranges of the stem circumference that were established and the
dimensions of the trees that were selected for instrumentation.
Table 2.1. Established circumference size classes (CSC) established, the number
of trees in the class (n), the sample tree representing each class,
sample tree circumference (Sc) and the fraction represented by each
sample tree in the ‘Delta’ Valencia orchard as a percentage (pt).
CSC (mm) n Sc (mm) Probe Depths
(mm)
pt
≤400 12 327 7, 14, 22, 30 17
401-449 34 382 7, 20, 30, 40 49
≥45 24 477 7, 20, 30, 40 34
Four probes were inserted in the sapwood of each tree at various depths from the
bark to capture the radial variation of sap flux density with sapwood depth. The
depths to which the different probes were inserted in each sample tree are shown in
Table 2.1. One measuring probe set consisted of a heater probe and two
thermocouples placed upstream and downstream at a distance of 50 mm (Figure
2.2). The 60-mm long line-heaters were made from 1.8-mm outside-diameter
stainless steel tubing, enclosing a constantan filament. Type T (copper/constantan)
thermocouples were embedded in 2 mm outside diameter polytetrafluoroethene
(PTFE) tubing. Drilling was performed using a drill guide strapped to the tree to
ensure that holes were parallel.
Core samples, with a diameter of 5 mm, were taken using an incremental borer from
trees that closely resembled the chosen trees. Simple measurements showed that
the bark thickness was 5 mm and sapwood depth of the trees was 40 mm. Sapwood
depth was based on wood discolouration. The core samples were preserved in
© University of Pretoria
46
formalin, acetic acid and alcohol and taken to the laboratory for anatomical studies of
the xylem structure.
Figure 2.2. Sap flow probes installed in two trees in the ‘Delta’ Valencia orchard.
Insert shows situation of the probes around a stem of one of the sample
trees.
The Vh for each probe set was calculated following Marshall (1958) as:
𝑉ℎ =𝑘
𝑥𝑙𝑛 (
𝑣1
𝑣2) × 3600 Equation 2. 1
where k is the thermal diffusivity of green (fresh) wood (assigned a nominal value of
2.5 x 10-3 cm2 s-1), x is distance in cm between the heater and either thermocouples
(= 0.5 cm), v1 and v2 are changes in temperature after the heat pulse is released
(from initial temperatures) as measured by the upstream and downstream
© University of Pretoria
47
thermocouples and 3600 converts seconds to hours. Heat pulse velocities were
measured and logged on an hourly basis using a CR10X data logger and an
AM16/32B multiplexer (Campbell Scientific Inc., Logan, Utah, USA).
2.2.2.2 Wounding correction
The heat pulse velocities were corrected for wounding (i.e. xylem tissue mechanical
damage) caused by installation of probes and heating using coefficients b, c, and d
obtained from a numerical model developed by Burgess et al. (2001) using the
following equation:
𝑉𝑐 = 𝑏𝑉ℎ + 𝑐𝑉ℎ2 + 𝑑𝑉ℎ
3 Equation 2.2
where Vc is the corrected heat pulse velocity. The functions describing the correction
coefficients in relation to wound width (w) are as follows:
𝑏 = 6.6155𝑤2 + 3.332𝑤 + 0.9236 Equation 2.3
𝑐 = −0.149𝑤2 + 0.0381𝑤 − 0.0036 Equation 2.4
𝑑 = 0.0335𝑤2 − 0.0095𝑤 + 0.0008 Equation 2.5
The wound width was initially assumed to be equal to the size of the largest drill bit
which was 0.2 cm for the thermocouples (Green and Clothier, 1988; Dragoni et al.,
2005). Visual inspection during the measurement period on a piece of wood
chiselled from one of the trees (Figure 2.3) and after Tree 1 was cut down at the end
of the measurement season (Figure 2.4) revealed a wounding width of 0.2 cm.
2.2.2.3 Determining sap velocity
Sap velocity (Vs) was calculated from the corrected heat pulse velocity using the
equation suggested by Marshall (1958) that was later modified by Barrett et al.
(1995):
𝑉𝑠 =𝑉𝑠𝜌𝑏(𝑐𝑤+𝑚𝑐𝑐𝑠)
𝜌𝑠𝑐𝑠 Equation 2.6
where ρb is the basic density of wood, cw and cs are specific heat capacity of the
wood matrix (1200 J kg-1°C-1 at 20 °C (Becker and Edwards, 1999) and sap (water,
4182 J kg-1 °C-1) at 20 °C (Lide, 1992), respectively, mc is water content of the
sapwood and ρs is the density of water (1000 kg m-3).
© University of Pretoria
48
Figure 2.4. Measurement of the wounding width on a wood sample cut
transversely from the trunk of one of the sample trees at the end of the
measuring season in August 2009 (photo courtesy of Mark Gush).
B A
Figure 2.3. Sample of wood taken
from one of the sample trees for the
determination of wounding width,
sapwood water content and
sapwood density: (A) the sample of
wood just before removal (B)
measurement of the wounding
width and (C) the wood sample
used for measurements
C
© University of Pretoria
49
The sapwood moisture content and basic density were determined on freshly
excised wood samples. The sapwood samples were chiselled from the tree and
immediately placed in a plastic bag to preserve moisture content. The green mass of
the sample was measured; the volume (cm³) was determined by water-displacement
in a beaker and then oven-dried at a constant temperature of 70 °C until there was
no significant change in mass. Sapwood moisture content was calculated as:
𝑚𝑐 =𝑀𝑤−𝑀𝑑
𝑀𝑤 Equation 2. 7
where Mw and Md are wet and dry mass (kg) of the sapwood sample.
The basic density ρb was calculated as:
𝜌𝑏 =𝑀𝑑
𝑉𝑤 Equation 2. 8
where Vw is the volume of the wood sample which is equal to the volume of water
displaced in the beaker.
2.2.2.4 Converting sap velocity to sap flux density
Volumetric flow for individual probes was calculated as the product of sap velocity
(Vs) and cross-sectional area of conducting sapwood represented by the specific.
Integration of sap flux measurements by individual probes to obtain whole stem flux
(Q) was performed following the method of weighted average of heat pulse velocity
with depth as presented by Hatton et al. (1990) using Equation 2.9.
𝑄 = 𝜋[𝑟12 ∗ 𝑣1 + (𝑟2
2 − 𝑟12) ∗ 𝑣2 + (𝑟3
2 − 𝑟22) ∗ 𝑣3 + (𝑟4
2 − 𝑟32) ∗ 𝑣4] Equation 2.9
where vk is the heat pulse velocity measured by sensor k placed between radii rk-1
and rk. Integrated sap flux (assumed to be equal to whole tree transpiration) over the
whole sapwood area, calculated on an hourly basis, was normalised to millimetres of
water by dividing by the area allocated to each tree in the orchard (15.125 m2). The
annular cross-section of the sapwood of the tree was divided into four concentric
annuli as shown in Figure 2.5.
© University of Pretoria
50
Figure 2.5. Idealised cross-sectional areas of four annuli rings represented by probe
sets inserted at different depths in the conducting sapwood of a tree
stem shaded in different colours.
2.2.2.5 Xylem anatomical assessments
Xylem anatomical assessments were conducted on excised stem wood samples
prior to probe insertion and at the conclusion of sap flow measurements in wood
tissue surrounding the drilled holes in order to determine the extent of wounding
caused by implantation of probes in the sapwood of the ‘Delta’ Valencia trees and to
assess the thermal homogeneity of the wood. Transverse sections above and below
the drilled holes, radial sections on both sides of the drilled hole and tangential
sections across the drilled hole were made using a sliding microtome. To aid with
contrast during wood tissue observation sections were stained with safranin and fast
green and observed with a Wild Leitz GMBH light microscope (Abbott Laboratories,
Illinois, USA). Safranin normally dyes lignin red and fast green appears a brilliant
Cambium
Sapwood depth
Centre of the stem
Probe 2
Probe 4
Probe 1
Probe 3
r4
r3
r1
r2
r2
© University of Pretoria
51
green in cytoplasm and cellulosic cell walls (Glime and Wagner, 2013). These stains
are considered generally good for staining plant tissues and hence are normally
used for wood. Photomicrographs of the xylem structure and vessel distribution were
taken with a PowerShot A630 digital camera (Canon Inc., USA) mounted on the
microscope. Average xylem vessel lumen diameter and distance between groups of
vessels were measured with an optical scale on the microscope. To get an overall
picture of the sapwood, individual photos of cross sections, taken in radial sequence,
were then ‘stitched’ together using HP Photosmart Essential software (Hewlett-
Packard Development Company, L.P.).
2.2.3 Measurement of orchard evapotranspiration
ETc measurements to calibrate the sap flow method were conducted in the orchard
during two window periods, from 31 July to 3 August 2008 (winter) and from 21 to 25
of May 2009 (autumn). ETc during the two window periods was measured using the
eddy covariance method by the Hydroscience Research Group of the Council for
Scientific and Industrial Research (CSIR) based in Stellenbosch. The measurement
of ETc during winter 2008 was conducted soon after installation of the Vh system and
the measurement in autumn coincided with the end of the orange fruit growing
season in the orchard. Vh measurements were done throughout the season in fruit
growing season in the orchard.
2.2.3.1 The eddy covariance system
Fluxes of latent (λETc) and sensible heat (H) were measured with an extended open
path eddy covariance (OPEC) system, comprising a Campbell CSAT3 three-
dimensional sonic anemometer and in autumn an LI-7500 open path infrared gas
analyser (LiCor Inc., Lincoln, Nebraska, USA) was used, which were mounted on a
lattice mast 6.2 m above the soil surface at the centre of the field. The fetch
requirements of approximately 300 m were met from all directions. The configuration
of instruments is shown in Figure 2.6. Measurements were sampled at a frequency
of 10 Hz and logged on a CR5000 data logger (Campbell Scientific Inc., Logan,
Utah, USA). Air temperature and humidity were measured using a Vaisala HMP45C
temperature and humidity probe (Vaisala Oyj, Vantaa, Finland). Net irradiance (Rn)
was measured using two net radiometers at 6.4 m above the soil surface (Model
© University of Pretoria
52
240-110 NR-Lite, Kipp and Zonen, Delft, The Netherlands), with one placed above
the trees and the other between the tree rows. Soil heat flux (G) was determined
from four soil heat flux plates (model HFT-S, REBS, Seattle, Washington, USA)
placed within the tree rows and in-between the tree rows at a depth of 80 mm. A
system of four Campbell TCAV-L soil temperature averaging probes at depths of 20
and 60 mm were used to calculate the heat stored above the plates. Volumetric soil
water content in the first 60 mm was measured using a Campbell CS616 time
domain reflectometer. These sensors were connected to a CR23X datalogger and
measurements were performed at 10 Hz frequency, with averages obtained every 30
minutes. In order to ensure that the ETc measured were representative of the
orchard a quality analysis of the high frequency data for the EC was done using
EdiRe software (Campbell Scientific Inc., Logan, Utah, USA). The rotations did not
change the fluxes that much for the Groblersdal site.
Energy balance closure during the winter period was assessed using the energy
balance ratio (EBR) on a daily basis and was calculated as follows:
𝐸𝐵𝑅 = 𝐻+𝜆𝐸𝑇𝑐
𝑅𝑛−𝐺 Equation 2.10
where H, λETc, Rn and G are expressed in MJ m-2 day-1. Throughout the May 2009
measurement period, when the IRGA was used to estimate LE, less than 10 % error
in energy balance measurements was found (i.e. EBR>0.9 or EBR<1.1). In August
2008 the shortened energy balance was used to estimate λETc from measurements
of H, Rn and G.
2.2.4 Calibration of the HRM sap flow technique
Transpiration measured with the sap flow method and up-scaled to orchard level was
compared to ETc measured using the eddy covariance method during the two
window periods i.e. from 1 to 4 August 2008 (winter) and from 21 to 25 of May 2009
(autumn). These periods were chosen because the orchard was expected to have
negligible rates of evaporation of intercepted water from the canopy or ground
surface (due to limited ground cover and the absence of rainfall) as observed in
© University of Pretoria
53
orchards by Rogers et al. (1983), Pernice et al. (2009) and Marin and Angelocci
(2011). The contribution of irrigation water to evaporation was also considered
negligible, since the orchard was drip-irrigated and water was applied directly under
the tree, wetting small areas of soil, which were shaded by the trees. As a result ETc
was expected to equal orchard transpiration.
Figure 2.6. Micro-meteorological instruments including mounted on a lattice mast at
the centre of the ‘Delta’ Valencia orchard (photo courtesy of Mark Gush).
A sensitivity analysis was performed to assess the effect of wound width, sapwood
water content and sapwood density on final transpiration values during the autumn
measurement period. The percentage change in transpiration was determined as a
result of a percentage change in each parameter, which varied between 5 and 30%,
whilst keeping the other parameters at their original values. Based on the results of
the sensitivity analysis and the fact that sapwood water content and density were
estimated using experimental procedures the calibration procedure was targeted on
wound width, as in the case of Zreik et al. (2003) and Fernández et al. (2006).
Net radiometers
Temperature & humidity
sensor
Sonic anemometer
IRGA
© University of Pretoria
54
A wound width termed a “virtual wound width” that could be used in the conversion of
heat pulse velocities to tree transpiration, so that average daily sap flow
measurements of the sample trees matched daily ETc measurements, was
determined using linear regression equations. Using this method, HRM-based
transpiration was initially calculated using different wound widths (2, 3, 4 and 5 mm).
Linear regression equations were then formulated between HRM-based transpiration
and the wound width. A “virtual wound width” that matched HRM-transpiration to ETc
was calculated by inputting the measured ETc into the regression equations. Daily
values were chosen to avoid issues associated with capacitance within the citrus
trees, which could result in time lags between sap flow and ETc measurements.
2.3 Results
2.3.1 Calibration of the HRM sap flow technique
It is evident that when using the observed wound width of 0.2 cm, up-scaled sap flow
underestimated the ETc measured over the orchard using the eddy covariance
method. This underestimation of transpiration during the two measurement periods
ranged from 50 to 192 %. A sensitivity analysis of wound width, sapwood water
content and sapwood density revealed that final transpiration values were in fact
most sensitive to changes in wound width, where a 30 % change in wound width
resulted in an error of 50 % in transpiration (Figure 2.7). Although final transpiration
values were also shown to be fairly sensitive to changes in sapwood density and
sapwood water content, a survey across citrus species revealed these two
parameters to be fairly conservative in well-irrigated citrus trees (Table 2.2). The use
of these conservative values for citrus is therefore likely to result in an error of less
than 13%, which is deemed acceptable.
© University of Pretoria
55
Figure 2.7. Error analysis illustrating resultant percentage error in transpiration from
the HRM due to errors in selected parameters in the up-scaling
equations from Vh’s to transpiration, whilst maintaining all other
parameters at their original value.
Table 2.2. Sapwood water content and sapwood density for various citrus species. A
total of 19 samples were taken, with six from lemon, nine from sweet
orange, two from soft citrus and two from grapefruit.
Parameter Value SD %CV
Sapwood water content (kg kg-1) 0.61 0.8 13.11
Sapwood density (kg m-3) 0.74 0.04 5.41
The slopes, intercepts and R2 values of the regression equations of HRM-
transpiration against wound width on different days are presented in Table 2.3. The
virtual wound widths that were determined by inputting measured ETc into the
regression equations are also shown in Table 2.3. The average virtual wound width
that resulted in good agreement between daily sap flow-based transpiration and daily
ETc measurements during the winter and autumn periods were 3.3 and 3.1 mm,
respectively. Figure 2.8 shows a very good agreement between daily transpiration
-60
-40
-20
0
20
40
60
-60 -50 -40 -30 -20 -10 0 10 20 30 40
Err
or
in t
ran
sp
ira
tio
n (
%)
Error in parameter (%)
Wound width
Sapwood density
Sapwood moisture content
© University of Pretoria
56
(obtained using a virtual wound width of 3.3 mm for the winter period and 3.1 mm for
the autumn period) and ETc.
Table 2.3. Slopes and intercepts of the regression equations used to determine
wounding width by inputting ETc on different days. The R2 values and the
virtual wound widths determined from the regression equations are also
shown.
Date Slope Intercept R2
value
Wound width
(mm)
Winter 2008
31 July 7.25 -0.41 0.99 3.4
1 August 6.79 -0.39 0.99 3.7
2 August 6.62 -0.23 0.99 3.1
3 August 8.08 -0.49 0.99 3.1
Average 3.3
Autumn 2009
21 May 3.83 -0.14 0.99 3.1
22 May 4.17 -0.17 0.99 2.8
23 May 3.62 -0,06 0.99 3.0
24 May 3.81 -0.06 0.99 3.3
25 May 5.67 -0.05 0.99 3.2
Average 3.1
There was a small difference of 2 mm between the average virtual wound widths
obtained in the winter and autumn measurement periods. The average of the two
virtual wound widths, i.e. 3.2 mm, was then used to convert heat pulse velocities to
tree transpiration for the two short-term seasonal crop ETc measurement campaigns.
A comparison of hourly transpiration obtained using the average virtual wound width
with measured ETc resulted in a good agreement between measured ETc and
transpiration from the calibrated HRM for most of the days considered (Figure 2.9).
Regression analysis between transpiration and ETc when forced through the origin
resulted in slopes of 1.21 and 1.07 in winter and autumn.
© University of Pretoria
57
Evapotranspiration (mm day-1
)
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Tra
nspiration (
mm
day
-1)
0.0
0.5
1.0
1.5
2.0
2.5
3.0
Winter
Autumn
Figure 2.8. Comparison between daily transpiration obtained using a virtual wound
width of 3.3 mm for the winter period and 3.1 mm for the autumn period.
2.3.3 Xylem anatomy
A ‘stitched’ transverse section of the citrus sapwood before the insertion of the
probes illustrates that citrus trees have diffuse–porous xylem anatomy i.e. early- and
latewood pores show little variation in size or density (Figure 2.10A). The xylem
elements occur in two to four radial groups or clusters, which form the xylem vessels.
Measured average lumen diameter of the xylem vessels was (0.15 ± 0.02) mm and
the average distance between clusters was (0.8 ± 0.22) mm. Figure 2.10B is a
photograph of the position of a heater probe in the sap wood of one of the
instrumented trees showing the extent of the damage caused by drilling and constant
heating. The average wound width that was observed for all the probes by the naked
eye was 0.2 cm. Tangential sections around the same hole showed blocked xylem
elements appearing as dark stripes as illustrated in Figure 2.10C. The discolouration
of ray cells and vessels blocked with gum (g) and phenols (p), that extend for at least
6 mm from the drilled holes (Figure 2.10D), which could also be seen a transverse
section of the sapwood approximately 1 mm above the downstream thermocouple
© University of Pretoria
58
(Figure 2.10E), suggests that the effective wounding caused by drilling and heating
extends beyond the visible wound width of 2 mm.
21-May-09 22-May-09 23-May-09 24-May-09 25-May-09 26-May-09
0.00
0.05
0.10
0.15
0.20
0.25
31-Jul-08 01-Aug-08 02-Aug-08 03-Aug-08 04-Aug-08
Evap
otr
anspir
ation
and
Tra
nspir
ation
(m
m h
-1)
0.0
0.1
0.2
0.3
0.4
Transpiration
Evapotranspiration
Date
Winter
Autumn
Figure 2.9. Up-scaled HRM-based transpiration using a “virtual wound width” of 3.2
mm as compared to micrometeorological measurements of
evapotranspiration.
© University of Pretoria
59
Figure 2.10. Photomicrographs of sapwood samples taken from a ‘Valencia’ orange
tree. (A) A transverse section of the citrus sapwood prior to probe
insertion showing the distribution of the xylem vessels (x) in the
sapwood; (B) a transverse section of sapwood adjacent to one of the
drilled holes made for insertion of a probe; (C) tangential section
adjacent to one of the drilled holes showing the affected rays; (D) a
transverse section showing the extent of blockage of xylem vessels by
gum (g) and phenols (p); and (E) a transverse section of the sapwood
approximately 1 mm above the downstream thermocouple showing
extensive xylem blockage. Scale bars on each photograph are equal to
2 mm.
A
C
D
E
Hole
Blocked xylem vessel
vessels
x
g p
Hole
B
© University of Pretoria
60
2.4 Discussion
Heat pulse velocity techniques have been used extensively to estimate transpiration
of fruit trees (Nicolas et al., 2005; Rana et al., 2005; Pernice et al., 2009; Marin and
Angelocci, 2011; Consoli and Papa, 2013). There is serious concern that this
technique tends to underestimate transpiration (Green and Clothier 1988; Jones et
al., 1988; Steppe et al., 2010), and species-specific calibration is therefore required
(e.g., Green and Clothier, 1988; Green et al., 2003; Dragoni et al., 2005; Ferandez et
al. 2006; Pernice et al., 2009). This is because of the invasive nature of the methods
that disrupts the sap stream and alters the thermal homogeneity of the sapwood
surrounding the probes causing a systematic underestimation in the measured Vh’s
(Cohen et al., 1981; Green and Clothier, 1988). In this study sap flow measurements
using the HRM were calibrated using eddy covariance measurements of ETc under
field conditions in an orchard planted with ‘Delta’ Valencia orange trees, when
evaporation was considered to be negligible.
The small difference between the virtual wound widths determined during the two
measurement periods used for calibration, which were nine months apart, indicates
that limited changes occurred in the sapwood surrounding the probes during the
measurements. As the initial calibration was performed immediately after the
insertion of the probes it can be inferred that wounding in response to probe insertion
in citrus occurs quickly and once established it does not change over time. As such a
single calibration using this method is most probably sufficient for long-term
measurement of transpiration using the HRM in citrus. In addition, the use of a virtual
wound width of 3.2 mm resulted in very good agreement between transpiration
determined by the HRM and ETc suggesting that this is a reliable method for in-field
calibration of sap flow systems. Constant wound widths or correction factors for long
term sap flow measurements have also previously been reported. Dragoni et al.
(2005) reported a constant wounding effect over a period of two and half months in
apples. Rana et al. (2005) also used a constant correction factor over a period of one
month in citrus trees. This is, however, in contradiction to earlier findings where
changes in wound width were observed to occur with time (Swanson and Whitfield,
1981; Marshall et al., 1981; Zreik et al., 2003).
© University of Pretoria
61
Although there are no reports of the calibration of the HRM in citrus in literature to
draw comparisons with, similar calibrations of other Vh techniques do exist. The
average virtual wound width of 3.2 mm that could be used to match orchard
transpiration to ETc for both measurement periods is similar to the range of values
(2.8-3.2 mm) determined by Zreik et al. (2003) in citrus. However, it is greater than
the virtual wound width of 2.4 mm determined in citrus by Fernández et al. (2006).
Differences would be expected due to the fact that experimental conditions and
analytical equations used by the authors are not the same as those used in this
study. This includes the fact that they were using a different sap flow method (i.e. the
compensation heat pulse technique) with different wound correction analytical
models. In addition both experiments were not carried out under field conditions but
were cut tree experiment (Fernández et al., 2006) and transpiration based on mass
changes of potted plants (Zreik et al., 2003).
There are two possible explanations that could be drawn from the xylem anatomical
assessments that could have led to a virtual wound width that is greater than what
was observed with the naked eye. Firstly, the diameter of the xylem vessels and the
distance between them is greater than 0.4 mm, which may cause the sapwood
matrix of the citrus trees to depart from the assumption of thermal homogeneity and
therefore may not adhere to the underlying principles of the Vh techniques as
theorised by Huber and Schmidt (1937). Studies by Swanson (1983), Green and
Clothier (1988) and Fernández et al. (2006) concluded that distances of non-
conductive tissue greater than 0.4 mm tend to cause departures from thermal
homogeneity such that an effective wound width greater than the largest hole drilled
would be required to convert Vh’s to transpiration. This is not taken into account in
the wound correction process, based on the numerical models of Burgess et al.
(2001) used in this study. Secondly, the greater virtual wound width is also probably
due to the fact that besides the interruption of flow pathways (caused by mechanical
damage during sensor installation) intact vessels had become occluded as the plant
responded to wounding by deposition of gum and phenols. According to Burgess et
al. (2001) such occlusions, which may not be visible to the naked eye, result in a
region of non-conducting wood around the site of probe insertion, which has an
effect of decreasing the heat ratio (v1/v2) and hence Vh in Equation 2.3.
© University of Pretoria
62
2.5 Conclusions
The HRM was successfully calibrated under field conditions in a citrus orchard using
ETc estimated from the eddy covariance method. It was established that a virtual
wound width of 3.2 mm can be used in the conversion of heat pulse velocities to tree
transpiration for both measurement periods (winter 2008 and autumn 2009) so that
average daily sap flow measurements of the sample trees matched daily ETc
measurements done using the eddy covariance method. The virtual wound width
that was greater than the biggest hole drilled (2 mm) was probably due to the fact
that citrus wood is not thermally homogeneous as indicated by the xylem anatomical
assessments. This in-field calibration method was performed twice, with nine months
of continuous heat pulse velocity measurements separating the measurement
windows, indicating the likelihood that the wound width remained constant
throughout this period. The use of a single virtual wound width during the two
measurement periods resulted in a good match between transpiration and ETc
thereby suggesting that the HRM is reliable for long-term measurements of
transpiration in citrus; and a single calibration using this method may be sufficient for
long-term transpiration measurements. The development of a constant “virtual”
wound width for citrus is a novel approach which will in future allow the online
calculation of the sap flux density for potential real time monitoring of citrus tree
transpiration.
© University of Pretoria
63
Chapter 3: Measurement of long-term transpiration and transpiration
coefficients in citrus orchards
3.1 Introduction
Irrigation plays a major role in meeting crop water requirements for evergreen citrus
orchards, which require water all year round in semi-arid regions, such as South
Africa (Bijzet, 2006). Micro-irrigation systems, such as drip, that allow for greater
control over the quantity of water applied, have been adopted by citrus farmers in
South Africa in recent years. The potential of drip irrigation systems to improve
irrigation water use efficiency in orchards lies in their ability to supply near-to-exact
the water required by the tree crops; and the fact that it applies water directly to the
root system on small areas which are usually shaded, thereby limiting evaporation
(Reinders et al., 2010)
However, there is a need to complement these systems with water management
procedures that are system-specific, if their benefits are to be fully realized.
Knowledge of the water use of crops and the factors that govern the actual process
of water uptake are essential for developing models that are used for extrapolating
such information for irrigation water management. Such knowledge is considered
lacking under the relatively new micro-irrigation systems in South Africa (Pavel et al.,
2003; Volschenk et al., 2003). The need to update such information is also
necessitated by a change in rootstock choice from the more vigorous ‘Rough Lemon’
to the rootstocks of intermediate vigour which include ‘Carrizo’, ‘Troyer’ and
‘Swingle’. Thermometric sap flow techniques can be utilised to expedite the
gathering of information on transpiration of citrus orchards to fill that knowledge gap.
The heat ratio method (HRM) sap flow technique has been used to study
transpiration dynamics in Jatropha curcas L. (Gush, 2008), olive (Williams et al.,
2004; Er-Raki et al., 2010), Eucalyptus marginata (Bleby et al., 2004) and Prosopis
(Dzikiti et al., 2013).
The crop coefficient (Kc) approach proposed by Doorenbos and Pruitt (1977) that
was later refined by Allen et al. (1998) has been used quite extensively in irrigation
water management of various crops including citrus (e.g., Green and Moreshet,
1979; Hoffman et al., 1982; Rogers et al., 1983; Castel et al., 1987; Castel, 1997;
© University of Pretoria
64
Allen et al., 1998; Yang et al., 2003; Rana et al., 2005; Fares et al., 2008). This has
been attributed to the relative simplicity of the method such that it is currently
considered the standard method for determining crop water use (Pereira et al.,
2015). One major advantage of the crop coefficient approach is that it considers the
independent contributions of soil evaporation and crop transpiration towards crop
evapotranspiration (ETc) (Allen et al., 1998). This is done by using the dual crop
coefficient approach in which Kc is split into two separate coefficients, i.e. a soil
evaporation coefficient (Ke) and the crop transpiration coefficient (Kcb).
Recent studies have, however, revealed variability in crop coefficients of citrus
among different growing regions (Carr, 2012; García Petillo and Castel, 2007) and
the need to up-date crop coefficients has been highlighted (Snyder and O’Connell,
2006). One source of variation that is particular to Kcb values that are based on
transpiration determined using sap flow methods (e.g., Rana et al. 2005; Villalobos
et al., 2009; Marin and Angelocci, 2011; Villalobos et al., 2013) is that the Kcb values
of Allen et al. (1998) represent the transpiration component of ETc and a small
evaporation component from soil that is visibly dry at the surface. Although the
evaporation component is small, it is not determined by the sap flow methods. To
distinguish the two forms, Villalobos et al. (2013) suggested that sap flow-based crop
coefficients should be termed transpiration coefficients (Kt).
The aims of this work were to: (i) quantify the long-term transpiration of three well-
irrigated citrus orchards using the HRM, (ii) determine Kt values of citrus orchards
based on transpiration measured using the HRM, and (ii) verify the suitability of the
Kcb values proposed for a generic citrus crop in the FAO56 to predict transpiration of
citrus orchards.
3.2 Materials and methods
3.2.1 Experimental site and climate
Measurement of tree water use was conducted at Moosrivier Farm (Schoeman
Boerdery Group) in commercial orchards planted with ‘Delta’ Valencia and
‘Bahianinha’ Navel oranges [Citrus sinensis (L.) Osbeck]. The description of the
© University of Pretoria
65
‘Delta’ Valencia orange tree orchard and equipment setup is given in Section 2.2.1.
Measurements were conducted during the 2008/9 season in the ‘Delta’ Valencia
orchard and during the 2009/10 in the ‘Bahianinha’ Navel orchard. The ‘Bahianinha’
Navel orange trees were grafted on ‘Carrizo’ rootstocks and were planted in 2003.
The row orientation was 51o ENE, with north being 0o. Tree spacing was 6 m x 2 m
with the trees planted on ridges, pruned to an average height of 2.5 m each year
after harvest. Measured leaf area index (LAI) of the orchard was 2.34 m2 m-2. The
orchard was drip-irrigated, with one line per tree row and pressure compensating
emitters with a discharge of 2.3 L h-1, spaced 1 m apart. The soil type was sandy
loam with an average of 5-10 % clay in the first 1 m depth.
Both orchards were managed according to commercial standards, as stipulated by
the GLOBALG.A.P standards. Irrigation scheduling was done using a simple water
balance approach checked with soil water content measurements performed using
DFM capacitance probes (DFM Software Solutions, South Africa). Generally, both
orchards received irrigation on a daily basis, with two to three 2-h pulses per day,
which ensured that soil water content was maintained near to field capacity.
An additional data set of transpiration and ETo measurements that was sourced from
within the “Water use of fruit tree orchards” project, of which this study was part of,
will also be reported on. The measurements were conducted in a commercial
orchard planted with ‘Rustenburg’ Navels in the 2010/11 season at Patrysberg Farm
in the Western Cape Province (32° 27’ 15.43’’ S and 18° 58’ 3.58’’ E, 149 m.a.s.l.)
near Citrusdal, in the winter rainfall region of South Africa. The area receives an
average annual rainfall of 200 mm and has average minimum and maximum
temperatures of 10 °C and 24 °C, respectively. The ‘Rustenburg’ Navel trees were
grafted on ‘Troyer’ citrange rootstocks and were planted in 1996. Tree spacing was 5
× 2.5 m with trees being pruned shortly after harvest to a height of 3.2 m, with
selective limb removal, according to the industry standards of the production area.
The orchard row orientation was 79° ENE. Average tree height was 3.3 m. Measured
leaf area index (LAI) of the orchard was 3.40 m2 m-2. The orchard was drip-irrigated,
with two drip lines per tree row using pressure compensating emitters spaced 0.8 m
apart with a discharge of 1.8 L h−1. The soil texture was a sandy loam with an
© University of Pretoria
66
average of 5–10 % clay in the top 1-m depth. The orchard was irrigated on a daily
basis, with one 2 to 3 h pulse per day.
3.2.2 Estimation of transpiration
Heat pulse velocities (Vh) were measured using the HRM developed by Burgess et
al. (2001) on three sample trees in the ‘Delta’ Valencia orchard, four sample trees in
the ‘Bahianinha’ Navel orchard and six trees in the ‘Rustenburg’ Navel. Details of the
instrumentation in the ‘Delta’ Valencia orchard were given in Section 2.2.2. Selection
of the sample trees that were instrumented in the other orchards was also based on
a tree stem circumference survey as described the ‘Delta’ Valencia orchard in
Section 2.2.2. Table 3.1 shows circumference size classes established, the number
of trees in each class, the sample tree representing each class, sample tree
circumference and the percentage of the orchard represented by each sample tree in
the ‘Bahianinha’ and ‘Rustenburg’ Navel orchards. No heartwood was detected on
core samples taken from other trees in three orchards using an incremental borer.
The details of the measurement procedure of the Vh’s and conversion to
transpiration volumes are given in Section 2.2.2. The integrated sap flux density
(assumed to be equal to transpiration) was calculated for every hour and then
summed to obtain daily transpiration. The virtual wound width that was established
for the ‘Delta’ Valencia orchard using micrometeorological measurements in Chapter
2 was used in the three orchards.
© University of Pretoria
67
Table 3.1. Established circumference size classes (CSC), the number of trees in the
class (N), the sample tree representing each class, sample tree
circumference (Sc) and the percentage represented by each sample tree
each orchard (Pt).
CSC (mm) N Sc (mm) Probe depths
(mm)
Pt
‘Bahianinha’ Navels
≤282.7 24 222 7, 14, 22, 30 16.00
282.8-306.3 26 306 7, 20, 30, 40 18.00
306.4-342.4 55 313 7, 20, 30, 45 37.00
≥342.5 42 346 7, 20, 30, 40 29.00
‘Rustenburg’ Navels
≤300 6 280 8, 15, 25, 30 21.10
300-350 15 345 8, 15, 25, 30 32.40
350-390 15 384 10, 18, 30, 40 8.45
350-390 11 392 10, 18, 30, 40 8.45
390-430 12 442 10, 18, 30, 40 21.10
>430 12 444 10, 18, 30, 40 8.50
Integrated volumetric sap flow of the individual trees (L day−1) was converted to
transpiration (mm day−1) using the ground area allocated to each tree in the orchard,
i.e. 12 m2 in the ‘Bahianinha’ Navel orchard and 12.5 m2 in the ‘Rustenburg’ Navel
orchard. Orchard transpiration in the two orchards was calculated as a weighted
average of sampled trees as suggested by Hultine et al. (2010), based on a stem
circumference survey at the start of the study and the consistent relationship
between seasonal water use and stem circumference (Figure 3.1). Transpiration of
the ‘Delta’ Valencia orchard was calculated as an average of the sample trees as a
result of the sandwich pruning method adopted in this orchard, which resulted in a
significantly smaller canopy in every second tree. In the ‘Bahianinha’ and
‘Rustenburg’ Navel orchards where trees were pruned to almost an equal canopy
size, it was thought that the stem diameter would have an influence on the
transpiration rate.
© University of Pretoria
68
Trunk circumference (cm)
0 10 20 30 40 50
Tota
l w
ate
r use (
mm
)
0
200
400
600
800
1000
'Rustenburg' Navels
'Bahianinha' Navels
y = 2.744x1.4507
R2 = 0.9963
y = 2.8608x1.4589
R2 = 0.9978
Figure 3.1. Relationship between trunk circumference and seasonal water use for
the ‘Bahianinha’ and ‘Rustenburg’ Navel orchards
3.2.3 Reference evapotranspiration
Weather variables required for the determination of reference evapotranspiration
(ETo) were recorded using an automatic weather stations (AWS) installed at both
sites. The weather variables that were recorded included wind speed, solar radiation,
temperature, relative humidity and rainfall. The AWS at Groblersdal was located on
an open stretch of mown, rain-fed grass and was 50 m west of natural vegetation,
which consisted of sparse trees (2-3 m tall) and grasslands. There were buildings to
the east, within 130 m of the AWS, and irrigated orchards within 300 m to the north.
Under these fairly dry conditions, reference evapotranspiration (ETo) is likely to be
slightly overestimated, as compared to the reference surface (Allen, 2008). This is as
a result of high VPD that occurs over the equilibrium boundary layer over a dry
surface as compared to a well water grass surface theorised by Allen et al. (1998).
The AWS at Citrusdal was located on the site of a recently top-worked (1 m height),
ridged and drip irrigated orchard, with a short ground cover consisting of grass and
weeds. It was 20 m east of an irrigated orchard with a height of 3 m, which would
© University of Pretoria
69
have reduced wind speed and therefore ETo is likely to be somewhat
underestimated, as compared to standard conditions (Allen 2008).
Quality assessment and quality control of the data was performed according to the
procedures described by Allen (2008). The “upper” measured values of solar
radiation (Rs) fell routinely below the clear-sky short wave radiation (Rso) curve and
Rs measurements were therefore adjusted upwards based on the average value of
Rs/Rso on clear days (Allen 2008).
Daily ETo was determined using the FAO-56 Penman-Monteith equation following
Allen et al. (1998):
𝐸𝑇𝑜 =0.408∆(𝑅𝑛−𝐺)+𝛾
900
𝑇𝑎+273𝑢2(𝑉𝑃𝐷)
∆+𝛾(1+0.34𝑢2) Equation 3.1
where ETo is reference evapotranspiration (mm day-1), Rn is net radiation at the crop
surface (MJ m-2 day-1), G is soil heat flux density (MJ m-2 day-1), Ta is mean daily air
temperature at 2 m height (°C), u2 is wind speed at 2 m height (m s-1), VPD is water
vapour pressure deficit (kPa), Δ is slope of the vapour pressure curve (kPa °C-1) and
γ is the psychrometric constant (kPa °C-1).
3.2.4 Determination of transpiration (Kt) crop coefficients
According to Villalobos et al. (2013) the basal coefficient of Allen et al. (1998)
includes some soil evaporation which makes it difficult to separate the two
components of ETc effectively. An equivalent of the basal crop coefficient, termed the
transpiration coefficient (Kt) as suggested by Villalobos et al. (2013), was determined
from measured tree transpiration and ETo as follows:
𝐾𝑡 =𝑇
𝐸𝑇𝑜 Equation 3.2
where Kt is the basal crop coefficient and T is transpiration estimated from sap flow
measurements. Monthly crop coefficients were calculated from monthly totals of
measured daily orchard transpiration and ETo. Monthly crop coefficients were
calculated as citrus is an evergreen crop and significant changes in canopy size do
not occur on a weekly basis.
© University of Pretoria
70
3.3 Results and discussion
3.3.1 Measurement of long-term transpiration
Measured transpiration rates in the three orchards showed large day to day
variations, which were largely determined by changes in atmospheric evaporative
demand, defined in terms of ETo (Figure 3.2). At Groblersdal the recorded ETo in the
2008/9 season ranged from 0.93 mm day-1 to 8.04 mm day-1. Transpiration in the
‘Delta’ Valencia orchard on the days when minimum and maximum ETo were
measured were 0.24 and 1.92 mm day-1 respectively. ETo in the 2009/10 season
ranged from 0.71 mm day-1 to 7.93 mm day-1. Transpiration of ‘Bahianinha’ Navel
orange trees on the days with minimum and maximum ETo were 0.08 and 1.63 mm
day-1, respectively. ETo in the 2010/11 season at Citrusdal ranged from 0.59 to 9.58
mm day-1 and; transpiration rates measured in the ‘Rustenburg’ Navel orchard on
those respective days were 0.26 and 2.16 mm day-1. Maximum transpiration rates
recorded in all the orchards were not on days with the maximum ETo.
Differences in transpiration measured in the orchards become more evident when
the average rates for the different seasons (i.e. summer and winter) are considered
(Table 3.2). Similar seasonal averages have also been calculated by other scientists
(Moreshet and Green, 1983; García Petillo and Castel, 2007) for easy comparison
with other data. Transpiration rates were generally higher in the ‘Delta’ Valencia
orchard than in the ‘Bahianinha’ Navel orchard despite the close similarities in the
atmospheric evaporative demand. This was largely attributable to the fact that the
‘Delta’ Valencia trees were bigger in size as compared to the ‘Bahianinha’ Navel
trees. Furthermore, although higher atmospheric evaporative demands were
observed in Citrusdal during the summer months (winter rainfall region), the
‘Rustenburg’ Navels did not use proportionally more water as compared to the ‘Delta’
Valencia orchard in Groblersdal (summer rainfall region), even though canopy size
was similar. This has previously been noted by Kaufmann (1977), who observed that
although evaporative demand was much higher under Arizona desert conditions, as
compared to the humid conditions in Florida, citrus transpiration was very similar in
summer in these two regions.
© University of Pretoria
71
30-J
ul-08
19-A
ug-0
8
08-S
ep-0
8
28-S
ep-0
8
18-O
ct-
08
07-N
ov-0
8
27-N
ov-0
8
17-D
ec-0
8
06-J
an-0
9
26-J
an-0
9
15-F
eb-0
9
07-M
ar-
09
27-M
ar-
09
16-A
pr-
09
06-M
ay-0
9
26-M
ay-0
9
15-J
un-0
9
05-J
ul-09
25-J
ul-09
Re
fere
nce
Eva
po
tra
nsp
ira
tio
n a
nd
Tra
nsp
ira
tio
n
(mm
da
y-1
)
0
2
4
6
8
10
Reference Evapotranspiration
Transpiration
01-S
ep-0
9
21-S
ep-0
9
11-O
ct-
09
31-O
ct-
09
20-N
ov-0
9
10-D
ec-0
9
30-D
ec-0
9
19-J
an-1
0
08-F
eb-1
0
28-F
eb-1
0
20-M
ar-
10
09-A
pr-
10
29-A
pr-
10
19-M
ay-1
0
08-J
un-1
0
28-J
un-1
0
18-J
ul-10
0
2
4
6
8
10
'Bahianinha' Navels
'Delta' Valencia
Date
13-A
ug-1
0
02-S
ep-1
0
22-S
ep-1
0
12-O
ct-
10
01-N
ov-1
0
21-N
ov-1
0
11-D
ec-1
0
31-D
ec-1
0
20-J
an-1
1
09-F
eb-1
1
01-M
ar-
11
21-M
ar-
11
10-A
pr-
11
30-A
pr-
11
20-M
ay-1
1
09-J
un-1
1
29-J
un-1
1
19-J
ul-11
08-A
ug-1
1
0
2
4
6
8
10
12
'Rustenburg' Navels
Figure 3.2. Daily transpiration and reference evapotranspiration determined in the
‘Delta’ Valencia, ‘Bahianinha’ and ‘Rustenburg’ Navel orchards.
© University of Pretoria
72
Table 3.2. Average daily transpiration (T) and reference evapotranspiration (ETo) for
the summer and winter seasons in the three orchards.
Orchard T (mm day-1) ETo (mm day-1)
Summer Winter Summer Winter
‘Delta’ Valencia 2.30 1.16 5.75 2.57
‘Bahianinha’
Navel 1.90 0.65 6.09 2.53
‘Rustenburg’
Navels 2.26 2.08 6.42 1.62
Published work on seasonal water use of citrus orchards under South African
conditions is available in the form of classic work done over two decades ago – a
knowledge gap that prompted this study. Transpiration of a basin-irrigated 15-year
old Valencia orchard measured by Green and Moreshet (1979) in weighing
lysimeters ranged from 1.9 to 5.8 mm day-1, with an average of 2 mm day-1 in winter
and 4.97 mm day-1 in summer. The reference evapotranspiration measured using a
class ‘A’ pan during the same period ranged between 2.5 and 7.6 mm day-1, which is
very similar to the environment in which this study was conducted. The seasonal
transpiration figures were higher than the values in the current study partly because
the trees were older and transpiration (mm) was calculated from the wetted area,
instead of the area allocated to the tree. If these values are normalised using the
area allocated to each tree, transpiration would range from 0.69 to 1.73 mm day-1
which is comparable to what was established in this study.
Daily transpiration measured in the three citrus orchards was within the range of ETc
(0.7 to 8.5 mm day-1) reported in citrus tree species across the different growing
regions of the world (García Petillo and Castel, 2007; Carr, 2012). However, there
are few studies in which long-term transpiration of citrus orchards was measured
using sap flow techniques in other regions that can be compared with this study.
Transpiration rates measured in this study were lower than 4 mm day-1 in summer
and 1 mm day-1 in winter reported by Rana et al. (2005) under Mediterranean
© University of Pretoria
73
conditions. The difference can most likely be ascribed to the manner in which
transpiration volumes (L) are converted to mm, as described above. Rana et al.
(2005) used the base of the tree crown to convert L to mm day-1, whilst in this study
the area allocated to each tree was used. Consoli and Papa (2013) also measured
transpiration rates greater (0.5 to 6 mm day-1) in semi-arid Mediterranean conditions
than what was observed in this study likely because the trees were older and had a
larger canopy (LAI ranging from 4.0 to 4.7) than those in this study.
Besides canopy size, an additional factor which may have contributed to the lower
transpiration in the ‘Delta’ Valencia and ‘Bahianinha’ orchards used in this study than
other orchards (including the ‘Rustenburg’ Navel orchard reported here), is the lower
than average yields attained during the monitoring periods. The yields for the ‘Delta’
Valencia and ‘Bahianinha’ orchards of 28 and 15 t ha-1, respectively, were
considerably lower than the average yields for these orchards and much lower than
the 75 t ha-1 recorded in ‘Rustenburg’ Navel orchard. There are reports of crop load
impacting gaseous exchange (Syvertsen et al., 2003), as well as sap flow rates
(Yonemoto et al., 2004), with lower crop yields having a negative feedback on
photosynthesis and stomatal conductance, through carbohydrate accumulation in
leaves. Differences in transpiration between Valencia and Navel trees may be due to
different varietal, as well as the rootstock-scion combinations as reported previously
(Yonemoto et al., 2004; Rodríguez-Gamir et al., 2010). However, despite all of these
factors transpiration rates measured in this study are comparable to 0.4 to 2.3 mm
day-1 that was reported by Marin and Angelocci (2011) in Southern Brazil, and 0.1 to
2.5 mm day-1 reported by Villalobos et al. (2013) in East and South Spain. It is
therefore evident that citrus transpiration, on both a daily and seasonal basis, varies
quite dramatically across the different growing regions of the world.
Transpiration was measured for periods of 364 days in the ‘Delta’ Valencia, 301 days
in the ‘Bahianinha’ Navels and 365 days in the ‘Rustenburg’ Navels. Total
transpiration for the measurement periods were 682 mm, 468 mm and 672 mm in
the ‘Delta’ Valencia, ‘Bahianinha’ and ‘Rustenburg’ Navel orchards, respectively
(Table 3.3). The total rainfall and irrigation volumes recorded for each orchard also
shown in Table 3.3 exceeded transpiration suggesting that the trees were not water
stressed. Irrigation management was assumed to be very good especially given the
© University of Pretoria
74
fact that these were the best growers producing export quality citrus. Rana et al.
(2005) measured a seasonal transpiration totalling 766 mm using sap flow, which is
higher than what was recorded in the ‘Delta’ Valencia orchard of similar age.
Although cumulative transpiration could not be ascertained in the cases of either
Marin and Angelocci (2011) or Villalobos et al. (2013) the close similarity of the daily
magnitudes would most likely translate to the same effect on a seasonal basis. Total
crop evapotranspiration, including evaporation in citrus orchards, has been
estimated to be in the range of 767 mm yr-1 in Uruguay (García Petillo and Castel,
2007) to over 1400 mm yr-1 in Florida USA (Fares et al., 2008).
Table 3.3. Accumulated reference evapotranspiration (ETo), rainfall, irrigation and
transpiration for the measurement periods in each orchard.
Accumulated
water (mm) ‘Delta’ Valencia
‘Bahianinha’
Navel
‘Rustenburg’
Navels
ETo 1668 1423 1528
Rainfall 579 518 203
Irrigation 925 261 600
Transpiration 682 468 672
3.3.2 Determination of transpiration crop coefficients (Kt)
There was a lot of variation in the day-to-day Kt coefficients determined in the three
orchards (Figure 3.3). The ranges for the Kt values determined were 0.14-0.78 in the
‘Delta’ Valencia orchard, 0.12-0.59 in the ‘Bahianinha’ Navel orchard and 0.22-1.21
in the ‘Rustenburg’ Navel orchard. Differences in the Kt values can be observed
when average monthly Kt (Figure 3.4) and seasonal Kt (Table 3.4) values were
determined. The monthly Kt coefficients were fairly constant in both summer and
winter for the orchards at Groblersdal, which is in agreement with other reports in
citrus (e.g., Kalma and Stanhill, 1969; Green and Moreshet, 1979; Allen et al., 1998).
The Kt values established for the ‘Rustenburg’ Navels are significantly higher in
winter than what was observed during the summer. Similar trends were reported by
García Petillo and Castel (2007) in Uruguay and Villalobos et al. (2013) in Spain.
© University of Pretoria
75
'Bahianinha' Navel
Date
01
-Se
p-0
9
01
-Oct-
09
31
-Oct-
09
30
-No
v-0
9
30
-De
c-0
9
29
-Ja
n-1
0
28
-Fe
b-1
0
30
-Mar-
10
29
-Ap
r-1
0
29
-May-1
0
28
-Ju
n-1
0
28
-Ju
l-1
0
Kt
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
'Delta' Valencia
31
-Ju
l-08
30-A
ug-0
8
29-S
ep-0
8
29
-Oct-
08
28
-No
v-0
8
28
-De
c-0
8
27
-Ja
n-0
9
26
-Fe
b-0
9
28
-Ma
r-0
9
27-A
pr-
09
27
-Ma
y-0
9
26
-Ju
n-0
9
26
-Ju
l-09
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
13
-Au
g-1
0
12
-Se
p-1
0
12
-Oct-
10
11
-No
v-1
0
11
-De
c-1
0
10
-Ja
n-1
1
09
-Fe
b-1
1
11
-Mar-
11
10
-Ap
r-1
1
10
-May-1
1
09
-Ju
n-1
1
09
-Ju
l-1
1
08
-Au
g-1
1
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4'Rustenburg' Navel
Figure 3.3. Transpiration coefficients (Kt) determined for the ‘Delta’ Valencia,
‘Bahianinha’ Navel and ‘Rustenburg’ Navel orchards.
© University of Pretoria
76
'Delta' Valencia
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Kt
Kcb (FAO56)
'Rustenburg' Navel
Month
Au
gu
st
Se
pte
mb
er
Octo
be
r
No
ve
mb
er
De
ce
mb
er
Ja
nu
ary
Fe
bru
ary
Ma
rch
Ap
ril
Ma
y
Ju
ne
Ju
ly
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
'Bahianinha' Navel
Kt
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Figure 3.4. Average monthly transpiration coefficients (Kt) for the three orchards.
Also shown are the FAO-56 standardised basal crop coefficients (Kcb
FAO-56) for a citrus orchard with 70 % canopy cover (‘Delta’ and
‘Rustenburg’ orchards) and 50 % ground cover (‘Bahianinha’ orchard),
all with no active ground cover, as given by Allen and Pereira (2009).
© University of Pretoria
77
Table 3.4. Average transpiration crop coefficients (Kt) and water vapour pressure
deficit (VPD) for the summer and winter seasons in the three orchards.
Orchard Kt VPD (kPa)
Summer Winter Summer Winter
‘Delta’ Valencia 0.43 0.38 1.37 1.16
‘Bahianinha’ Navel 0.35 0.28 1.47 0.98
‘Rustenburg’ Navel 0.31 0.71 2.24 0.74
The most obvious difference in Kt values between orchards was found in the winter
months, where values were considerably higher in the winter rainfall region, which is
most likely attributable to the lower average vapour pressure deficit (VPD) (Table
3.4) and higher average transpiration rates at this time in this region (Table 3.2).
Kriedemann and Barrs (1981) and Oguntunde et al. (2007) both reported that VPD is
the dominant regulator of transpiration in citrus, when trees are well watered, with
transpiration decreasing with an increase in VPD. The lower VPD was also reflected
in lower ETo in winter (Table 3.2) in the winter rainfall region than in the summer
rainfall region.
A logical explanation for the trend of Kt values in the three orchards is that as ETo
increases during the summer months, tree transpiration does not increase at the
same rate. Figure 3.2 suggests that the transpiration measured in both orchards
does not always increase in proportion to the increase in ETo. This suggests that
there is a maximum transpiration rate for these citrus orchards, which is reached
when ETo exceeds approximately 5 mm day-1. The maximum transpiration rate for
the ‘Delta’ Valencia orchard is approximately 3 mm day-1, in the ‘Bahianinha’ Navels
approximately 2 mm day-1 and approximately 2.5 mm day-1 in the ‘Rustenburg’
Navels. The limited response to high ETo values is more apparent when one
considers the sharp responses that occur when ETo falls to low values in the three
orchards (Figure 3.2). The limited responses to high ETo is despite the fact that trees
© University of Pretoria
78
are actively growing (in terms of flushes) and they have growing fruit on them. This is
suggestive of the fact that citrus trees might only be able to supply a certain
maximum amount of water to the canopy, irrespective of the atmospheric demand,
as highlighted by Sinclair and Allen (1982).
Kriedemann and Barrs (1981) suggested that this could partly be explained by the
low root:shoot ratio of citrus and the poorly developed root hairs that are not capable
of supplying enough water to match atmospheric demand during the peak periods.
Rodriguez-Gamir et al. (2010) working in young Valencia orange trees also
confirmed that transpiration of citrus tree species is strongly influenced by the leaf to
root ratio. This regulation mechanism has also been associated with stomatal
closure under conditions of high atmospheric evaporative demand (Kaufmann, 1977;
Kriedemann and Barrs 1981; Sinclair and Allen, 1982; García Petillo and Castel,
2007).
The Kt values obtained in the three orchards are much lower than the standardised
values reported by Allen and Pereira (2009) (Figure 3.4). If standard values
published by Allen et al. (1998) were to be applied over the measurement period in
each orchard, transpiration would have been overestimated by 94 % in the ‘Delta’
Valencia orchard, 98 % in the ‘Bahianinha’ Navel orchard and 96 % in the
‘Rustenburg’ Navel orchard. The ranges of seasonal Kt coefficients in the three
orchards compare very well to the range of 0.2-0.7 by Villalobos et al. (2013).
3.4 Conclusions
Average daily transpiration in the three orchards ranged from 0.77 mm day-1 to 2.26
mm day-1 depending on canopy size, season and climate (summer versus winter
rainfall areas). Total seasonal transpiration for the two orchards in which
measurements covered an entire production season, i.e. ‘Delta’ Valencia and
‘Rustenburg’ Navel orchards were 682 mm and 672 mm, respectively. Transpiration
coefficients derived from the measured data were lower than the standardised
values published in the FAO56. Differences in Kt values obtained amongst the three
orchards, as well as compared to values from other growing regions, means that this
kind of information is of significance to the commercial growers on whose farms the
© University of Pretoria
79
research was conducted. Such information is not always applicable and readily
transferrable to many other regions within South Africa or across citrus growing
regions of the world, or even for different seasons. This emphasizes the need to
develop a relatively simple method for estimating site-specific crop coefficients, in
order to improve water management. Sap flow-based measurements of transpiration
obtained in this study present a valuable data set for developing models of citrus
transpiration that can be used to extrapolate measured water use to other growing
regions of the world.
© University of Pretoria
80
Chapter 4: Estimation of transpiration crop coefficients from ground cover and
height in citrus orchards
4.1 Introduction
Knowledge of crop water use is the fundamental basis for efficient irrigation water
management strategies. The term irrigation water management encompasses
activities such as irrigation systems planning, irrigation scheduling and issuing of
water rights/permits. Measurements undertaken to gather information of crop water
use are time consuming, expensive and require specialised expertise, which is not
always available. The high variability between different orchards and the perennial
nature of orchards exacerbates the difficulty in obtaining water use estimates for
orchard crops (Hoffman et al., 1982; Castel et al., 1987). Consequently a number of
crop water use models have been developed which have proved very useful in
extrapolating measured data and predicting crop water use under a range of
conditions.
The use of the dual crop factor approach (Allen et al., 1998) has proved very useful
in micro-irrigated orchards, with discontinuous canopies, as it allows the separate
estimation of transpiration and evaporation. However, in tree crops, a linear
relationship between the evapotranspiration from a short, smooth and uniform grass
surface and a tall, very rough, clustered orchard canopy may not always hold true
(Annandale and Stockle, 1994; Testi et al., 2004). This probably explains the large
variation in Kt values determined among the three citrus orchards in this study
(Chapter 3), as well as crop coefficients (Kc and Kcb) reported in other citrus orchards
across the different growing regions of the world (see Table 1.3). This means that Kc
values derived in one location may not be readily transferable to other locations,
which limits their extrapolation to different climatic zones, with different orchard
management practices.
In an attempt to make crop coefficients more transferrable between different
orchards, Allen and Pereira (2009) developed a procedure for estimating crop
coefficients where vegetation density and height varies between orchards. These
authors found that under such conditions, Kc and basal crop coefficient (Kcb) values
© University of Pretoria
81
can be estimated more accurately by taking into account fraction of ground covered
or shaded by the vegetation, the height of the vegetation and the degree of stomatal
regulation under wet soil conditions. An adjustment can also be made for climate by
taking into account the average relative humidity and wind speed of the climate,
which accounts for the differences in roughness between the grass reference
surface and the tall orchard canopy. According to the authors the leaf resistance
term that accounts for stomatal regulation requires further investigation in order to
improve the accuracy of the method in orchards. The aim of this chapter was to
evaluate this procedure for the derivation of orchard specific Kt values in three citrus
orchards in different climatic regions of South Africa (summer and winter rainfall
regions), by comparing derived Kt values with actual Kt values determined from
transpiration measurements using sap flow. It was hypothesized that a dynamic
estimate of leaf resistance is needed to predict daily and seasonal changes of
transpiration in citrus orchards.
4.2 Materials and Methods
The long-term transpiration measured in the ‘Delta’ Valencia, ‘Bahianinha’ and
‘Rustenburg’ Navel orchards was used together with weather data that was
measured simultaneously at the site to determine transpiration crop coefficients (Kt).
Transpiration crop coefficients, as suggested by Villalobos et al. (2013), were chosen
as Kcb includes some evaporation when the soil surface is dry and only transpiration
was measured in this study. Details of the three orchards and the measurement of
transpiration are given in Section 2.2.4.1 and 3.2.2. Measurement of weather
variables and determination of reference evapotranspiration (ETo) were presented in
Section 3.2.3.
4.2.1 Model description
Allen and Pereira (2009) published a method of estimating orchard specific crop
coefficients from measurements of effective ground cover and tree height.
In drip-irrigated orchards the Kt coefficients, which correlate with transpiration of
planted trees, can be expressed in terms of a density coefficient (Kd) as:
© University of Pretoria
82
𝐾𝑡 = 𝐾𝑐 𝑚𝑖𝑛 + 𝐾d(𝐾t full − 𝐾𝑐 𝑚𝑖𝑛) Equation 4.1
where Kt is an approximation for transpiration crop coefficients for conditions
represented by the density coefficient, Kd; Kt full is the transpiration coefficient during
peak plant growth for conditions having nearly full ground cover (or LAI > 3) and Kc
min is the minimum crop coefficient for bare soil. The Kc min was set to zero, as only
transpiration from the trees measured using the heat ratio method sap flow
technique was considered.
The basal crop coefficients calculated following this approach apply to standard
conditions as stipulated in the FAO56 publication. As such it is recommended that
crop coefficients be adjusted for climates with minimum relative humidity (RHmin)
greater than or less than 45 % and/or with mean wind speeds at 2 m over grass (u2)
that are more or less than 2.0 m s-1. According to Allen et al. (1998) for large stand
size (greater than about 500 m2), Kt full for use with ETo can be approximated as a
function of mean plant height and adjusted for climate as:
𝐾t full = 𝐹r (min(1.0 + 0.1ℎ, 1.20) + [0.04(𝑢2 − 2) − 0.004(𝑅𝐻min − 45)] (ℎ
3)
0.3
)
Equation 4.2
where Fr [0-1] is a relative adjustment factor for stomatal control and h is the mean
maximum plant height (m) during the mid-season period, or full cover period.
Parameter Fr applies a downward adjustment (Fr ≤ 1.0) if the vegetation exhibits
more stomatal control on transpiration than is typical of most annual agricultural
crops, a situation that is typical of citrus (Kriedemann and Barrs, 1981).
Allen and Pereira (2009) suggested the following calculation for Fr for full cover
vegetation, based on the FAO Penman-Monteith equation and assuming full cover
conditions:
𝐹𝑟 ≈∆+𝛾(1+0.34𝑢2)
∆+𝛾(1+0.34𝑢2𝑟𝑙
100) Equation 4.3
where rl is the mean leaf resistance (s m-1); ∆ is the slope of the saturation vapour
pressure versus air temperature curve (kPa °C-1) and γ is the psychrometric constant
(kPa °C-1).
© University of Pretoria
83
The density factor (Kd) was determined according to Allen and Pereira (2009) as
follows:
𝐾d = min (1, 𝑀𝐿𝑓c eff, 𝑓c eff
(1
1+ℎ)) Equation 4.4
where ƒceff is the effective fraction of ground covered or shaded by vegetation [0.01-
1] near solar noon, ML is a multiplier of ƒceff describing the effect of canopy density
on shading and on maximum relative evapotranspiration per fraction of ground
shaded [1.5-2.0], with a value of 1.5 recommended for citrus (Allen and Pereirra,
2009).
Effective fraction cover was calculated as the ratio of tree canopy width to inter-row
spacing or the ratio of ground shaded area by the crop at solar noon to the total area
available to the tree, following Allen et al. (1998) in the ‘Delta’ Valencia orchard with
a north-south orientation. In the ‘Bahianinha’ and ‘Rustenburg’ Navel orchards,
provision had to be made for row orientation. In these orchards, ƒc eff was calculated
according to Allen et al. (1998) as follows:
ƒ𝑐 eff =ƒ𝑐
sin (𝛽) ≤ 1 Equation 4.5
where ƒc is the observed fraction of soil surface that is covered by vegetation as
seen from directly overhead. ƒc eff is usually calculated at solar noon, such that β
(mean elevation angle of the sun above the horizon during the period of maximum
evapotranspiration) can be calculated as:
𝛽 = arcsin [sin (𝜑)sin (𝛿) + cos (𝜑)cos (𝛿)] Equation 4. 6
where φ is latitude and δ is solar declination in radians. The average ƒc eff values
determined during the measurement period were 0.63, 0.61 and 0.88 for the ‘Delta’
Valencia, ‘Bahianinha’ and ‘Rustenburg’ Navel orchards, respectively.
4.2.2 Model calibration
Model calibration consisted of establishing parameters which provided for minimal
differences from measured Kt values on a monthly basis in all the orchards. Initially
the parameters recommended by Allen and Pereira (2009) for a generic citrus
© University of Pretoria
84
orchard were used to estimate Kt values in the study orchards. Average monthly
values of rl were then estimated by inverting Equation 4.4, after solving for Fr by
inverting Equation 4.3, using known monthly values of Kt full. Average monthly rl
values were determined based on the daily values and used to estimate Kt.
Measurements of leaf resistance in the ‘Rustenburg’ Navel orchard were performed
with a SC-1 leaf porometer (Decagon Device Inc, Pullman, WA, USA) on 5 sunlit
leaves per tree instrumented with sap flow equipment. Measurements were made
every hour from sunrise to sunset for a minimum of 3 days and an average was
obtained for each measurement period.
4.2.3 Model validation
Monthly Kt coefficients developed using the different approaches were used to
estimate transpiration in each orchard. The performance of the model was tested by
comparing the predicted and observed transpiration values. The statistical
parameters used to test the goodness of fit for the model were the coefficient of
determination (R2), root mean square error (RMSE) and mean absolute error (MAE)
(de Jager, 1994). The total error of estimating water use over the season was also
considered.
4.3 Results and discussion
Transpiration coefficients determined using the parameters given by Allen and
Pereira (2009) (Kt Allen and Pereira) were higher than the measured Kt values in the three
orchards (Figure 4.1). The overestimation of transpiration coefficients and crop water
use using the given citrus parameters is likely a reflection of the greater stomatal
control of transpiration in citrus than in most other crops, which is attributed to high
resistances to water transport within the plant (van Bavel et al. 1967; Kriedemann
and Barrs 1981; Sinclair and Allen 1982). Whilst Allen and Pereira (2009) account for
this by including their Fr parameter, which is used as a downward adjustment on
crop coefficients for crops which exhibit more stomatal control on transpiration than
most other agricultural crops, the rl value of 420 s m−1 suggested by the authors may
be too low, especially during hot summer months, when VPD increases.
© University of Pretoria
85
'Delta' Valencia
0.2
0.4
0.6
0.8
1.0
1.2
1.4 Kt measured
Kcb FAO56
Kt Allen & Pereira
Kt monthly avg. rl
Kt rl(ETo)
'Rustenburg' Navel
Month
Au
gu
st
Se
pte
mb
er
Octo
be
r
No
ve
mb
er
De
ce
mb
er
Ja
nu
ary
Fe
bru
ary
Ma
rch
Ap
ril
Ma
y
Jun
e
Ju
ly
0.2
0.4
0.6
0.8
1.0
'Bahianinha' Navel
Kt
0.2
0.4
0.6
0.8
1.0
Figure 4.1. Derived monthly transpiration crop coefficients (Kt) for the three
orchards. Transpiration crop coefficients were determined using
measured transpiration (Kt measured), the method described in Allen and
Pereira (2009) using the parameters given for citrus (Kt Allen and Pereira),
using estimates of monthly average rl values from transpiration data (Kt
monthly avg. rl) and using rl estimated from the relationship with ETo (Kt rl
(ETo)). Also shown are FAO-56 standardised basal crop coefficients
(Kcb FAO-56) for a citrus orchard with 70% canopy cover (‘Delta’ and
‘Rustenburg’ orchards) and 50% ground cover (‘Bahianinha’ orchard),
all with no active ground cover, as given by Allen and Pereira (2009)
© University of Pretoria
86
More appropriate values for rl were therefore estimated following a procedure
recommended by Allen and Pereira (2009), using measured transpiration data.
Following their approach rl values were estimated by inverting Equation 4.3, after
solving for Fr by inverting Equation 4.2 using known values of Kt full. Monthly, Kt full
values were obtained by scaling measured monthly Kt values with the appropriate Kd
value and were used to calculate a mean monthly rl (Figure 4.2). There was a very
good agreement between measured Kt values (Kt measured) estimated using the
calculated rl values (Kt monthly avg. rl) (Figure 4.1) with the two plots sitting perfectly on
top of one another. The perfect similarity between the measured and estimated Kt
values is expected since there is a high level of circularity involved in the estimation
process.
Month
August
Septe
mber
Octo
ber
Novem
ber
Decem
ber
January
Febru
ary
Marc
h
April
May
June
July
Me
an
lea
f re
sis
tance
(s m
-1)
0
1000
2000
3000
4000
'Delta' Valencia
'Bahianinha' Navels
'Rustenburg' Navels
Average
Allen and Pereira (2009)
Stomatal resistance measured in 'Rustenburg' Navels
Figure 4.2. Monthly mean leaf resistances calculated following the procedure
outlined by Allen and Pereira (2009), compared with the values
suggested by Allen and Pereira (2009) for citrus and daily stomatal
resistance measure in the ‘Rustenburg’ Navel orchard in Citrusdal.
© University of Pretoria
87
It is the magnitude and trend of rl values estimated following this method suggested
by Allen and Pereira (2009) that are of interest in as far as this result is concerned.
The rl values estimated in the three orchards ranged from 419 to 2 694 s m-1, which
is higher than the values suggested by Allen and Pereira (2009). Although Allen and
Pereira (2009) emphasised that the values of rl obtained using this procedure should
only be used in the estimation of Fr, these values were not too dissimilar to published
stomatal resistance data for citrus. Stomatal resistances in Shamouti orange varied
between 500 and 2000 s m-1 (Cohen and Cohen 1983); in young Navel orange trees
they varied between 200 to 8280 s m-1 (Dzikiti et al., 2007) and in oranges 295 and
830 s m-1 (Pérez-Pérez et al., 2008). The rl values in Figure 4.2 may therefore be
reasonable, and may not be heavily biased by the procedure outlined by Allen and
Pereira (2009), indicating that measured leaf resistances could potentially be used to
derive crop coefficients. However, average daily stomatal resistance, measured in
the various citrus orchards in Citrusdal, was between 300 and 1 000 s m−1 (Figure
4.2), which was considerably lower than the rl values of between 530 and 2,694 s
m−1 for this period, but generally the trend was very similar. Contrary to a constant
value of 420 s m-1, suggested by Allen and Pereira (2009), the average rl values
increased during the summer months in all three orchards. The increase in rl values
during the summer months, which are characterised by high atmospheric
evaporative demands, is evidence of strong stomatal control of transpiration in citrus.
The accurate estimation of Kt values and subsequent prediction of transpiration in
the three orchards using the method of Allen and Pereira (2009) hinged on the back-
calculation of rl. It is the estimation of mean rl, without measured transpiration, that
really hinders the ease with which this approach can be used to accurately estimate
crop coefficients for different citrus orchards. An independent methodology for
estimating this parameter was therefore sought by the parameterization of a
relationship between rl and routinely measured weather variables such as relative
humidity, VPD or ETo. A simple reproducible relationship with an excellent coefficient
of determination (R2 = 0.97) was found for the Citrusdal site during the 2010/11
season between rl and ETo (Figure 4.3).
© University of Pretoria
88
Reference evapotranspiration (mm day-1
)
0 2 4 6 8 10
Mean leaf re
sis
tance (
s m
-1)
0
500
1000
1500
2000
2500
3000
y = 316x - 61
R2 = 0.97
Figure 4.3. Relationship between reference evapotranspiration (ETo) and mean leaf
resistance for the ‘Rustenburg’ Navel orchard
Monthly transpiration estimates obtained in all the orchards, using the relationship
between ETo and rl, are presented in Figure 4.4. In the ‘Delta’ Valencia and
‘Bahianinha’ orchards, transpiration was underestimated at the beginning of the
season and overestimated at the end of the season. The underestimation at the start
of the season was more pronounced in the ‘Delta’ Valencia orchard, whilst the
overestimation at the end of the season was more pronounced in the ‘Bahianinha’
Navel orchard. In the ‘Rustenburg’ Navel orchard, in the 2010/11 season, good
estimates of water use were obtained throughout the season. The performance of
the model, as determined by statistical parameters, was good in the orchard in
Citrusdal, as the MAE was <20 % and D greater than 0.8. It did not, however,
perform as well in the orchards in the summer rainfall region with an MAE of 23 and
24 % and D of 0.55 and 0.63.
© University of Pretoria
89
'Delta' Valencia
0
20
40
60
80
100
Measured
Estimated
'Bahianinha' Navel MAE = 23 %RMSE = 14.9 mm
D = 0.55
Mo
nth
ly t
ransp
ira
tio
n (
mm
)
0
10
20
30
40
50
60
70
'Rustenburg' Navel
Month
Au
gu
st
Se
pte
mb
er
Octo
ber
No
ve
mb
er
De
ce
mb
er
Ja
nua
ry
Fe
bru
ary
Ma
rch
Ap
ril
Ma
y
Ju
ne
Ju
ly
0
20
40
60
80MAE = 5 %
RMSE = 3.7 mmD = 0.99
MAE = 24 %RMSE = 15.7 mm
D = 0.63
Figure 4.4. Monthly measured and estimated transpiration using Kt values derived
from mean leaf resistance, estimated from the relationship between
reference evapotranspiration and mean leaf resistance in the three
orchards. Statistical parameters shown include mean absolute error
(MAE), root of the mean square error (RMSE) and index of agreement
(D)
© University of Pretoria
90
Cumulative transpiration for the measurement periods was underestimated by 0.1 %
in the ‘Rustenburg’ Navels, and overestimated by 9 % in the ‘Delta’ Valencia orchard
and 18 % in the ‘Bahianinha’ Navel orchard when using the estimated Kt values (Kt rl
(ETo) (Figure 4.5). Given the empirical nature, the performance was fairly good,
especially compared to using the parameters stipulated by Allen and Pereira (2009)
for citrus. If derived Kt values using the parameters stipulated by Allen and Pereira
(2009), were to be applied over the measurement periods, they would have resulted
in a 139 % over-estimation of transpiration in the ‘Delta’ Valencia orchard, a 172 %
over-estimation in the ‘Bahianinha’ Navel orchard and a 95 % over-estimation in the
‘Rustenburg’ Navel orchard (Figure 4.5).
The accuracy of using the method of Allen and Pereira (2009) coupled with a
relationship between rl and ETo provided seasonal estimates within an acceptable
error range. This method could therefore be used for irrigation planning purposes
and for the issuing of water permits. However, the inability to predict water use
accurately on a monthly basis for most of the season limits the use of this procedure
for irrigation scheduling, which will require more reliable estimates of rl. This is not
unexpected, as stomatal conductance is known to be regulated by a number of
factors, which includes solar irradiance, ambient CO2 concentration, leaf to air water
vapour pressure deficit, leaf temperature, leaf water status and hydraulic limitations
to leaf water supply (Jarvis 1976; Sperry et al., 2002), and it is usually a combination
of these factors that determines stomatal conductance. This creates uncertainty
when only using climatic data to predict stomatal responses. In this respect, reduced
sink activity or accumulation of carbohydrates in leaves could also lead to decreases
in stomatal conductance (Iglesias et al., 2002; Syvertsen et al., 2003; Duan et al.
2008), which may be a contributing factor to the increased leaf resistance in the
Groblersdal orchards at the end of the season, as both these orchards experienced
lower than average yields during the monitoring period. As a result of the reduced
number of fruit on trees, and therefore sink strength, feedback inhibition of
photosynthesis through carbohydrate accumulation in leaves may have occurred.
Although Nebauer et al. (2013) attributed the lack of difference in photosynthetic
rates between high and low-yielding citrus trees, in alternate bearing cycles, to an
unsaturable sink in the root system of perennial fruit trees, their evidence is not
© University of Pretoria
91
compelling enough to suggest that changes in fruit load have no impact on overall
tree sink demand and canopy conductance. Further research is therefore required to
determine the relationship between yield and water use, as this has implications for
predicting seasonal water use. The solution for better prediction of leaf resistances
may be to use more mechanistic models which can predict canopy conductance
based on canopy size, environmental variables and sink strength or potential yield.
Similar approaches have been undertaken by Oguntunde et al. (2007) and Villalobos
et al. (2009, 2013) and have shown some promise, but the ability of these models to
predict citrus water use need to be verified in orchards with different canopy sizes
and in different climatic regions.
© University of Pretoria
92
'Delta' Valencia
0
200
400
600
800
1000
1200
1400
1600
1800
2000
ETo
Transpiration (Allen and Pereira)
Measured transpiration
Transpiration Kt rl (ETo)
'Bahianinha' Navel
Cum
mula
tive E
To
and T
ranspiration (
mm
)
0
200
400
600
800
1000
1200
1400
'Rustenburg' Navel
Date
August
Septe
mber
Octo
ber
Novem
ber
Decem
ber
Jan
uary
Feb
ruary
Marc
h
April
May
Jun
e
July
August
0
200
400
600
800
1000
1200
1400
1600
1800
Figure 4.5. Cumulative ETo, measured transpiration and transpiration estimated
from the derived crop coefficients in the ‘Delta’ Valencia orange orchard
(1 August 2008 to 31 July 2009), the ‘Bahianinha’ Navel orchard (3
September 2009 to 30 June 2010) and the ‘Rustenburg Navel orchard
(13 August 2010 to 12 August 2011)
© University of Pretoria
93
4.4 Conclusions
The development of transpiration coefficients from ground cover and crop height
published by Allen and Pereira (2009) was evaluated. Due to the greater stomatal
control of transpiration in citrus than in most other crops, crop coefficients that are
constant or are based on parameters that are constant are not suitable for accurate
predictions of water use, as crop coefficient models assume atmospheric demand-
limited conditions and transpiration in citrus is often water-supply limited, even in
well-watered orchards. Good agreement between measured and predicted
transpiration was obtained by using leaf resistances back-calculated from the
measured transpiration coefficient values. A relationship between monthly reference
evapotranspiration and mean leaf resistance provided a means of estimating a
dynamic leaf resistance which gives crop coefficients with a reasonable degree of
accuracy.
© University of Pretoria
94
Chapter 5: Predicting transpiration of citrus orchards using the Penman-
Monteith equation coupled with a Jarvis-type multiplication canopy
conductance model
5.1 Introduction
Transpiration coefficients determined according to Allen and Pereira (2009), based
on ground cover and crop height and using a dynamic estimate of leaf resistance,
provided good seasonal estimates of transpiration, but could not provide reasonable
monthly estimates throughout the season (Chapter 4). The leaf resistances
calculated using the measured transpiration data were higher than that suggested by
Allen and Pereira (2009) for citrus. This reiterates the strong stomatal control over
transpiration that has been reported in citrus (Kriedemann and Barrs, 1981; Meyer
and Green, 1981; Sinclair and Allen, 1982). A mechanistic approach that
incorporates the intra-annual dynamics of canopy conductance may therefore hold
the solution for improved modelling and/or simulation of citrus transpiration
(Villalobos et al., 2013; Taylor et al., 2015).
When measurements of transpiration are available, the bulk canopy conductance of
a vegetated surface can be obtained using the top-down approach by inversion of
the Penman-Monteith equation (Baldocchi et al., 1991). Such data sets have been
used to develop canopy conductance models, as a function of environmental
variables, for predicting transpiration using the Penman-Monteith equation (e.g.,
Granier and Loustau 1994; Green and McNaughton 1997; Zhang et al., 1997;
Oguntunde et al., 2007; Villalobos et al., 2009; Villalobos et al., 2013). Jarvis-type
models, i.e. multiplicative models of similar form to that originally proposed by Jarvis
(1976), have been parameterised using weather variables to model the variations in
canopy conductance with satisfactory accuracy in a number fruit trees e.g., olives
(Villalobos et al., 2000; Testi et al., 2006), cashew (Oguntunde and van de Giesen,
2005) and citrus (Cohen and Cohen, 1983; Oguntunde et al., 2007).
Besides being based on strong physiological relationships between stomatal
responses and environmental variables, the other advantage of the Jarvis model
(Jarvis, 1976) is that weather data available from a standard automatic weather
© University of Pretoria
95
station is used to predict canopy conductance. This approach therefore does not
require any additional input data than what is required for the FAO56 approach.
Whilst the use of the Penman-Monteith equation coupled with a Jarvis-type
conductance model by Oguntunde et al. (2007) in citrus showed some promise, it
was only evaluated for a short period of time in a single orchard. Whitley et al. (2009)
showed that the Penman-Monteith equation coupled with the Jarvis model with a
single set of parameters was able to accurately predict transpiration in a forest
species over an annual cycle. The ability of this approach, to estimate long-term
transpiration in citrus and its transferability amongst orchards with varying canopy
sizes and in different climatic regions, still needs to be verified.
The aims of this study were to: (i) Use sap flow-based transpiration measurements to
parameterise a Jarvis-type conductance model for inclusion in the Penman-Monteith
equation for predicting long-term transpiration of citrus orchards and (ii) To test the
transferability of the parameters of a Jarvis-type model obtained in an orchard in the
summer rainfall area to predict transpiration in orchards situated in the summer and
winter rainfall areas of South Africa. The hypothesis that a dynamic estimate of
canopy conductance is needed to predict daily and seasonal changes of
transpiration in citrus orchards was further tested.
5.2 Materials and methods
Long-term orchard transpiration and weather data measurements that were
conducted in the ‘Delta’ Valencia, ‘Bahianinha’ and ‘Rustenburg’ Navel orchards
were used in the modelling approach conducted in this chapter. Details of the
measurement of long-term transpiration and weather parameters are given in
Chapters 2 and 3. The weather parameters used included wind speed, solar
radiation, air temperature and relative humidity.
5.2.1 Calculation of canopy conductance
Bulk canopy conductance (gc) was calculated from transpiration measurements
using the inversion of the following Penman-Monteith equation (Monteith and
Unsworth, 1990):
© University of Pretoria
96
𝜆𝐸𝑐 =∆(𝑅𝑛−𝐺)+𝜌𝑎𝐶𝑝𝑔𝑎𝑉𝑃𝐷
∆+𝛾(1+𝑔𝑎𝑔𝑐
) Equation 5. 1
where λ is the latent heat of vaporization of water (J kg-1), Ec is canopy transpiration
(kg m-2 s-1), Δ is slope of the water vapour pressure curve versus temperature (kPa
K-1), Rn is net irradiance at the crop surface (W m-2), G is soil heat flux (W m-2), taken
as 10 % of Rn, ρa is the density of dry air (kg m-3), Cp is the specific heat capacity of
the air (J kg-1 K-1), VPD is saturation vapour pressure deficit (kPa), γ is the
psychrometric constant (kPa K-1), ga is the aerodynamic conductance (m s-1) and gc
is the canopy conductance (m s-1). Rn was estimated from solar irradiance measured
at the automatic weather stations according to Allen et al. (1998) using
measurements of citrus reflection coefficient obtained using an albedometer in the
‘Delta’ Valencia orchard.
Leaf boundary conductance (ga) was calculated from an empirical relation derived by
Landsberg and Powell (1973), which accounts for the mutual sheltering of clustered
leaves as:
𝑔𝑎 = 0.172𝑝−0.56 (𝑢
𝑑)
0.5
Equation 5. 2
where d is a characteristic leaf dimension, equal to 8 mm for citrus leaves (Dzikiti et
al., 2011), u is the mean wind speed (m s-l ) across the leaf surface, and p is a
measure of foliage density to wind given by the ratio of total leaf plane area to the
area of the foliage projected onto a vertical plane. Leaf area was determined using a
leaf area density of 2.5 m2 m-3 and tree canopy volume was determined according to
a cone shape following Annandale et al. (2003).
5.2.2 Modelling canopy conductance
Canopy conductance was modelled using a Jarvis-type multiplicative model (Jarvis
1976), similar to the one used by Oguntunde et al. (2007), on an hourly basis with
weather data as follows:
𝑔𝑐,𝑗 = 𝑔𝑐 𝑚𝑎𝑥ƒ(𝑆𝑅)ƒ(𝑉𝑃𝐷)ƒ(𝑇) Equation 5.3
where gc,j is the canopy conductance predicted by the Jarvis model, gc max is the
maximum canopy conductance (m s-1), ƒ(SR) is a function of solar radiation, ƒ(VPD)
is a function of water vapour pressure deficit and ƒ(Ta) is a function of air
© University of Pretoria
97
temperature. The functions have values ranging between 0 and 1. A response
function for soil water content has been included in the Jarvis-type model in some
studies, particularly native forests (e.g. Whitley et al. 2008), but as the orchards in
this study were well-irrigated this function was set to one. The control functions of air
temperature and solar irradiance were similar to those of Oguntunde et al. (2007)
and took the following forms:
ƒ(𝑆𝑅) =𝑆𝑅
𝑅𝑚(
𝑅𝑚+𝑘𝑅
𝑆𝑅+𝑘𝑅) Equation 5. 4
ƒ(𝑇𝑎) =(𝑇𝑎−𝑇𝐿)(𝑇𝐻−𝑇𝑎)𝑡
(𝑘𝑇−𝑇𝐿)(𝑇𝐻−𝑘𝑇) Equation 5.5
𝑡 =𝑇𝐻−𝑘𝑇
𝑘𝑇−𝑇𝐿 Equation 5.6
where kR and kT are model parameters for the respective functions in which they are
used, TL and TH are the lower and upper temperature limits to transpiration fixed at 0
and 45 oC, respectively (Wright et al., 1995; Oguntunde et al., 2007). Rm is an
arbitrary radiation constant, often fixed at 1000 W m-2 (e.g. Wright et al., 1995;
Sommer et al., 2002). For the control function for water vapour pressure deficit the
equation derived by Zhang et al. (1997) was used. The equation is stated as:
ƒ(𝑉𝑃𝐷) =1+𝑘𝐷1𝑉𝑃𝐷
1−𝑘𝐷2𝑉𝑃𝐷 Equation 5.7
where kD1 and kD2 are model parameters.
5.2.2.1 Model parameterisation
Parameters gc max, kR, kT, kD1 and kD2 were optimised by minimising the sum of
squares of the residuals of the measured and modelled canopy conductance as:
𝑆2(𝑘) = ∑ (𝑔𝑐𝑖 − 𝑔𝑐𝑗(𝑘, 𝑥𝑖))𝑛𝑖=1 Equation 5. 8
where gci is the ith value of canopy conductance calculated using transpiration data
by inverting Equation 5.1, gc,j is the corresponding canopy conductance value
predicted by the Jarvis model, k represents the model parameters (kR, kT, kD1 and
kD2) and xi is the input variables of the ith model value. Minimisation of S2 was carried
© University of Pretoria
98
out by optimising k using the Marquardt and Simplex iterative procedures in the
ModelMaker 3rd Edition software package (Cherwell Scientific Publishing Ltd, UK).
Days with the highest gc values (10 to 15 December 2008) were used.
5.2.2.2 Model validation
Validation of the model was performed by calculating canopy conductance using the
optimised parameters of the Jarvis model and using these values in the Penman-
Monteith equation to estimate hourly transpiration. Only transpiration values for the
day-time (0800 h to 1700h) period were used to evaluate the performance of the
model. These values were compared to the day-time transpiration measured using
the HRM technique from16 December 2008 to 26 June 2009, to assess the ability of
the model to predict long-term transpiration in the ‘Delta’ Valencia orchard.
The transferability of the parameters optimised in the ‘Delta’ Valencia orchard was
evaluated in a ‘Bahianinha’ Navel orchard at the same location and in a ‘Rustenburg’
Navel orchard in the winter rainfall area of South Africa. Maximum canopy
conductance (gc max) required in Equation 5.3 was calculated for these orchards
according to Whitehead (1998), which was also used by Whitely et al. (2009):
𝑔𝑐 𝑚𝑎𝑥 = 𝐿𝐴𝐼 × 𝑔𝑠 𝑚𝑎𝑥 Equation 5. 9
where LAI is leaf area index and gs max is the maximum stomatal conductance. LAI in
the three orchards was measured using an AccuPAR Ceptometer (model LP-80
Decagon Devices, Inc, Pulman, WA, USA). The gc max, estimated through the
optimisation process based on the gc data calculated by inverting the Penman-
Monteith equation in the ‘Delta’ Valencia orchard, was used to determine gs max for
citrus using the LAI of the orchard. The response parameters of the weather
variables (i.e. kR, kT, kD1 and kD2) were applied in their original forms, as determined
for the ‘Delta’ Valencia orchard.
5.2.3 Statistical analysis
The statistical parameters used to test the predicted values against the observed
data for the model are linear regression models, the coefficient of determination (R2),
root mean square error (RMSE), mean absolute error (MAE) and the index of
© University of Pretoria
99
agreement of Willmott (D) (de Jager, 1994). The model reliability criteria
recommended by de Jager (1994) are that R2 should be greater than 0.5, D should
be greater than 0.8 and MAE should be less than 20 %.
5.3 Results
5.3.1 Canopy conductance
A typical diurnal trend of gc in the ‘Delta’ Valencia orchard, calculated from the
inversion of the Penman-Monteith equation on a clear sky day (27 January 2009),
and weather variables are presented in Figure 5.1. The gc on this particular day
reached a maximum value of 5.7 m s-1 in the morning (0800 to 1100 h) and gradually
decreased to a minimum value of 1.9 mm s-1 at 1500 h. The high gc in the morning is
in response to an increase in solar irradiance, which causes a sudden increase in
the VPD on the evaporating leaf surface; whilst later on in the day the responses of
gc are controlled by a combination of variables VPD, SR, and Ta. This trend of high
and low values of gc in the early morning and late afternoon, which was observed on
most clear-sky days, is consistent with measurements of stomatal conductance of
citrus (Cohen and Cohen, 1983).
© University of Pretoria
100
00:00 04:00 08:00 12:00 16:00 20:00 00:00
VP
D (
KP
a)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
Time
00:00 04:00 08:00 12:00 16:00 20:00 00:00
gc (
mm
s-1
)
0
1
2
3
4
5
6
Time
00:00 04:00 08:00 12:00 16:00 20:00 00:00
Ta (
oC
)
18
20
22
24
26
28
30
00:00 04:00 08:00 12:00 16:00 20:00 00:00
SR (
W m
-2)
0
200
400
600
800
1000
1200
A
B
C
D
Figure 5.1. Typical diurnal trend of A) solar irradiance (SR), B) air temperature (T),
C) vapour pressure deficit (VPD) and D) canopy conductance (gc),
calculated from the inversion of the Penman-Monteith, in the ‘Delta’
Valencia orchard on a clear sky day, 27 January 2010.
Considerable scatter, signifying the stomatal response to weather variables, is
exhibited by the gc values when plotted over the measurement period in each
orchard (Figure 5.2). The gc values show a similar trend in the ‘Delta’ Valencia and
‘Bahianinha’ Navel orchards, where the highest values were observed in summer
and lowest values were at the end of the winter months. The highest gc values
recorded in the ‘Delta’ Valencia and ‘Bahianinha’ Navel orchards were approximately
12 mm s-1 and 9 mm s-1, respectively. In contrast, maximum canopy conductance
(10 mm s-1) in the ‘Rustenburg’ Navels orchard remained relatively constant
throughout the measurement period.
© University of Pretoria
101
Figure 5.2. Hourly canopy conductance (gc) calculated using the measured
transpiration and inverting the Penman-Monteith equation in the three
orchards. Gaps in the plotted data were due to missing hourly weather
data caused by power failure to the weather stations.
'Delta' Valencia
'Bahianinha' Navel
'Rustenburg' Navel
© University of Pretoria
102
5.3.2 Parameterisation of the Jarvis model
The five parameters, i.e. gc max, kD1, kD2, kT and kR, which describe the seasonal
response of canopy conductance to SR, VPD and Ta, obtained in the ‘Delta’ Valencia
orchard through the Marquardt and Simplex iterative procedures on an hourly basis,
are shown in Table 5.1. A very good correlation, R2 value of 0.97, was observed
between the measured and predicted canopy conductance during the optimisation
process in the ‘Delta’ Valencia orchard (Table 5.1). The three functions of SR, VPD
and Ta therefore explained over 97 % of the variation in canopy conductance in the
orchard and all parameter values were statistically significant (P<0.001). Figure 5.3
shows the normalised canopy conductance (gc/gc max) plotted against the functions of
weather variables (i.e. solar radiation; VPD and temperature) for the orchard. The
functional forms of the equations fit very well to the boundary regions described by
the data (Figure 5.3).
Table 5.1. Optimised parameters for the Jarvis model used to predict canopy
conductance with water vapour pressure deficit, solar irradiance and air
temperature as response functions in the ‘Delta’ Valencia orchard.
Parameter Value
gcmax (mm s-1) 14.0380
kD1 (kPa) 0.0477
kD2 (kPa) -0.1799
kT (oC) 18.6410
kR (W m-2) 350.1700
R2 0.9700
P value <0.0010
© University of Pretoria
103
B
Water vapour pressure deficit (kPa)
0 1 2 3 4 5
gc/g
c m
ax
0.0
0.2
0.4
0.6
0.8
1.0
1.2
A
Solar irradiance (W m-2
)
0 200 400 600 800 1000 1200
0.0
0.2
0.4
0.6
0.8
1.0
1.2
C
Air temperature (oC)
0 10 20 30 40
0.0
0.2
0.4
0.6
0.8
1.0
1.2
Figure 5.3. Normalised canopy conductance in the ‘Delta’ Valencia orchard plotted
against (A) solar radiation (B) vapour pressure deficit and (C)
temperature. The forms of the model functions, based upon the
optimised parameters, are shown as solid lines.
© University of Pretoria
104
5.3.2 Predicting transpiration using the Penman-Monteith equation coupled
with the Jarvis-type model
Figure 5.4 shows the relation between the gc predicted by the Jarvis model and gc
obtained from sap flow data by inverting the Penman-Monteith equation. The
predicted values were scattered around the 1:1 line with a reasonable correlation (R2
= 0.56). There is high variance between predicted and measured values at higher
values of gc. Despite the disparities between measured and estimated hourly gc,
when using these values in the Penman-Monteith equation fairly good predictions of
the observed range of daily transpiration volumes were obtained (Figure 5.5A). Both
the day-to-day variation in transpiration and the gradual decline daily transpiration
rates as winter approached were captured by the model. Regression analysis
resulted in a slope of close to 1 (y = 1.13) between predicted and measured
transpiration in the ‘Delta’ Valencia orchard (Figure 5.5B) and the statistical
parameters were within the ranges stipulated for good model performance by de
Jager (1994).
y = 0.7783x
R2 = 0.56
Measured gc (mm s
-1)
0 2 4 6 8 10 12 14 16 18
Pre
dic
ted g
c (
mm
s-1
)
0
2
4
6
8
10
12
14
16
18
1:1 line
Figure 5.4. Relationship between hourly gc predicted by the Jarvis model and gc
obtained from sap flow data by inverting the Penman-Monteith equation
in the ‘Delta’ Valencia orchard.
© University of Pretoria
105
A
Date
01
-De
c-0
8
01
-Ja
n-0
9
01
-Fe
b-0
9
01
-Ma
r-0
9
01
-Ap
r-0
9
01
-Ma
y-0
9
01
-Ju
n-0
9
01
-Ju
l-0
9
Tra
nspiration (
mm
day
-1)
0.0
0.5
1.0
1.5
2.0
2.5
Predicted
Measured
B
y = 1.13x
R2 = 0.69
MAE = 19.37 %
D = 0.89
RMSE = 0.09
0.0 0.5 1.0 1.5 2.0 2.5
Pre
dic
ted
tra
nsp
ira
tio
n (
mm
da
y-1
)
0.0
0.5
1.0
1.5
2.0
Measured transpiration (mm day-1
)
Figure 5.5. Measured and predicted daily transpiration in the ‘Delta’ Valencia
orchard for A) the period 16 December 2008 to 26 June 2009 and (B)
regression analysis. Statistical parameters include slope of regression
equation, coefficient of determination (R2), mean absolute error (MAE),
Willmot coefficient of agreement (D) and root mean square error
(RMSE). The missing data were due to missing hourly weather data
caused by power failure to the weather station.
© University of Pretoria
106
Whilst the model performed very well in the ‘Delta’ Valencia parameterisation
orchard using independent validation data, it is important that the model is
transferable and can be used to predict long-term transpiration in a wide range of
citrus orchards. The transferability of the parameters for the Jarvis-type canopy
conductance model was therefore evaluated in a ‘Bahianinha’ Navel orchard at the
same location and in a ‘Rustenburg’ Navel orchard in the winter rainfall region of
South Africa. The value for gs max for the ‘Delta’ Valencia orchard was 4.92 mm s-1,
and was used together with orchard LAI to determine gc max in the ‘Bahianinha’ and
‘Rustenburg’ Navel orchards using Equation 5.9. Following this approach hourly gc
was poorly estimated in both of these orchards (Figure 5.6), with an R2 value of 0.47
for the ‘Bahianinha’ Navel orchard (Figure 5.6A) and 0.41 ‘Rustenburg’ Navel
orchard (Figure 5.6B). In the ‘Bahaininha’ Navel orchard gc tended to be
overestimated (slope = 1.34), whilst in the ‘Rustenburg’ Navel orchard gc tended to
be underestimated (slope = 0.64). As a result transpiration was overestimated in the
‘Bahianinha’ Navel orchard throughout the measurement period and underestimated
in the ‘Rustenburg’ Navel orchard (Figure 5.7). Three of the four statistical
parameters (i.e. MAE, D and RMSE) indicated that the parameters for the Jarvis-
type model determined in the ‘Delta’ Valencia orchard did not predict transpiration
adequately in both the ‘Bahianinha’ and ‘Rustenburg’ Navel orchards (Figure 5.8).
© University of Pretoria
107
'Bahianinha' Navel
1:1 line
0 2 4 6 8 10 12 14
0
2
4
6
8
10
12
14
'Rustenburg' Navel
Measured gc (mm s
-1)
0 2 4 6 8 10 12
Pre
dic
ted
gc (
mm
s-1
)
0
2
4
6
8
10
12
y = 1.34x
R2 = 0.47
y = 0.64x
R2 = 0.41
1:1 line
Figure 5.6. Relationship between gc predicted by the Jarvis model using
parameters of the ‘Delta’ Valencia orchard and gc obtained from sap
flow data by inverting the Penman-Monteith equation in the
‘Bahianinha’ and ‘Rustenburg’ Navel orchards.
© University of Pretoria
108
'Rustenburg' Navel
Date
01
-Au
g-1
0
05
-Se
p-1
0
10
-Oct-
10
14
-No
v-1
0
19
-De
c-1
0
23
-Ja
n-1
1
27
-Fe
b-1
1
03
-Ap
r-1
1
08
-May-1
1
12
-Ju
n-1
1
17
-Ju
l-1
1
21
-Au
g-1
1
0.0
0.5
1.0
1.5
2.0
2.5
'Bahianinha' Navel
01
-De
c-0
9
21
-De
c-0
9
10
-Ja
n-1
0
30
-Ja
n-1
0
19
-Fe
b-1
0
11
-Mar-
10
31
-Mar-
10
20
-Ap
r-1
0
10
-May-1
0
30
-May-1
0
19
-Ju
n-1
0
Tra
nsp
ira
tio
n (
mm
)
0.0
0.5
1.0
1.5
2.0
Measured
Predicted
Figure 5.7. Time series of measured and predicted transpiration in the ‘Bahianinha’
and ‘Rustenburg’ Navel orchards using Jarvis parameters optimised in
the ‘Delta’ Valencia orchard.
© University of Pretoria
109
'Rustenburg' Navel
Measured transpiration (mm)
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Pre
dic
ted
tra
nsp
ira
tio
n (
mm
)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
y = 0.8123x
R2 = 0.59
MAE = 38.64D = 0.69
RMSE = 0.36
'Bahianinha' Navel
y = 1.2304x
R2 = 0.57
MAE = 28.79D = 0.73
RMSE = 0.33
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
Figure 5.8. Regression analysis between measured and predicted transpiration in
the ‘Bahianinha’ and ‘Rustenburg’ Navel orchard using parameters
determined in the ‘Delta’ Valencia orchard. Statistical parameters used
include slope of regression equation, coefficient of determination (R2),
mean absolute error (MAE), Willmot coefficient of agreement (D) and
root mean square error (RMSE).
As transpiration was poorly predicted in the ‘Bahianinha’ and ‘Rustenburg’ Navel
orchards when using the parameters for the ‘Delta’ Valencia orchard, it was decided
to parameterise the Jarvis model in the ‘Bahianinha’ and ‘Rustenburg’ Navel
orchards following the same procedure as in the ‘Delta’ Valencia orchard.
Parameterisation of the Jarvis model was done using data measured during the
© University of Pretoria
110
period of 3 to 9 December 2009 in the ‘Bahianinha’ Navel orchard; and 13 to 17
August 2010 in the ‘Rustenburg’ Navel orchard. A very good correlation (R2 >0.9)
was once again observed between the measured and predicted canopy conductance
during parameterisation, such that over 90 % of the variation in canopy conductance
in the two orchards was explained by the three functions of SR, VPD and Ta (Table
2). All parameter values were found to be statistically significant (P<0.001). Most of
the normalised canopy conductance data points were below the ideal situation
described by the functional forms in both the ‘Bahianinha’ Navel (Figure 5.9) and
Rustenburg Navel orchards (Figure 5.10).
Table 5.2. Parameters for the response functions of water vapour pressure deficit,
solar irradiance and air temperature for the Jarvis model determined
through the optimisation process in the ‘Bahianinha’ and ‘Rustenburg’
Navel orchards.
Parameter ‘Bahianinha’
Navel
‘Rustenburg’
Navel
gc max (mm s-1) 12.2970 18.0260
kD1 (kPa) 0.0503 0.0502
kD2 (kPa) -0.3083 -0.2673
kT (oC) 18.4871 19.3670
kR (W m-2) 363.2290 356.6990
R2 0.9900 0.9200
P value <0.0010 <0.0010
There were close similarities between the optimum parameters of the Jarvis-type
model obtained in the ‘Bahianinha’ and ‘Rustenburg’ Navel orchards (Table 5.2) and
those of the ‘Delta’ Valencia orchard (Table 5.1). Maximum differences amongst the
kD1, kD2, kT and kR parameters in respect of the ‘Delta’ Valencia orchard parameters
for the three orchards were 8.6, 33.9, 4.8 and 11.2 %, respectively. There were also
differences in the values of gc max that were obtained during the optimisation process
(Table 5.2) and those calculated using Equation 5.9 for the two orchards. The gc max
values obtained for the different orchards seem to reflect the sizes of the tree
canopies as they increase with measured LAI.
© University of Pretoria
111
Temperature (oC)
0 10 20 30 40
0.0
0.2
0.4
0.6
0.8
1.0
1.2
Solar radiation (W m-2)
0 200 400 600 800 1000 1200
0.0
0.2
0.4
0.6
0.8
1.0
1.2
Vapour pressure deficit (kPa)
0 1 2 3 4 5 6
gc/g
c m
ax
0.0
0.2
0.4
0.6
0.8
1.0
1.2
A
B
C
Figure 5.9. Normalised canopy conductance in the ‘Bahianinha’ Navel orchard
plotted against (A) solar radiation (B) water vapour pressure deficit and
(C) air temperature. The forms of the model functions based upon the
optimised parameters are shown as solid lines.
© University of Pretoria
112
B
Vapour pressure deficit (kPa)
0 2 4 6 8
gc/g
c m
ax
0.0
0.2
0.4
0.6
0.8
1.0
1.2
C
Temperature (oC)
0 10 20 30 40
0.0
0.2
0.4
0.6
0.8
1.0
1.2
A
Solar radiation (W m-2)0 200 400 600 800
0.0
0.2
0.4
0.6
0.8
1.0
1.2
Figure 5.10. Normalised canopy conductance in the ‘Rustenburg’ Navel orchard
plotted against (A) solar radiation, (B) water vapour pressure deficit
and (C) air temperature. The forms of the model functions based upon
the optimised parameters are shown as solid lines.
© University of Pretoria
113
Improved correlations were observed between gc predicted by the Jarvis-type model,
using orchard specific parameters (Table 5.2) and measured gc in the ‘Bahianinha’
and ‘Rustenburg’ Navel orchards (Figure 5.11). While there was a good agreement
between predicted and measured transpiration in the ‘Bahianinha’ Navel orchard
using parameters for the Jarvis model specific to this orchard, there were periods of
underestimation of transpiration in the ‘Rustenburg’ Navel orchard during spring of
2010 (August to October) and winter 2011 (April to August) (Figure 5.12).
'Bahianinha' Navel
0 2 4 6 8 10 12 14
0
2
4
6
8
10
12
14
y = 0.8916X
R2 = 0.60
'Rustenburg' Navel
Measured gc (mm s
-1)
0 2 4 6 8 10 12 14 16 18
Pre
dic
ted
gc (
mm
s-1
)
0
2
4
6
8
10
12
14
16
18
y = 0.8797x
R2 = 0.57
1:1 line
1:1 line
Figure 5.11. Relationship between hourly gc predicted by the Jarvis model and gc
obtained from sap flow data by inverting the Penman-Monteith equation
in the ‘Bahianinha’ and ‘Rustenburg’ Navel orchards.
© University of Pretoria
114
'Bahianinha' Navel
01
-De
c-0
9
21
-De
c-0
9
10
-Ja
n-1
0
30
-Ja
n-1
0
19-F
eb-1
0
11
-Ma
r-1
0
31
-Ma
r-1
0
20
-Ap
r-1
0
10
-Ma
y-1
0
30
-Ma
y-1
0
19
-Ju
n-1
0
Tra
nsp
iratio
n (
mm
)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
Predicted
Measured
'Rustenburg' Navel
Date
01
-Aug
-10
31
-Aug
-10
30
-Sep
-10
30
-Oct-
10
29
-Nov-1
0
29
-Dec-1
0
28
-Jan
-11
27
-Feb-1
1
29
-Mar-
11
28
-Apr-
11
28
-May-1
1
27
-Jun
-11
27
-Jul-11
26
-Aug
-11
0.0
0.5
1.0
1.5
2.0
2.5
3.0
Figure 5.12. Time series plot of measured and predicted transpiration in the
‘Bahianinha’ and ‘Rustenburg’ Navel orchards. The gaps in the plotted
data were due to missing hourly weather data caused by power failure
to the weather stations.
Regression analysis revealed reasonable linear relationships between predicted and
measured transpiration in both the ‘Bahianinha’ and ‘Rustenburg’ Navel orchards
(Figure 5.13). The model predicted transpiration reasonably well as almost all the
statistical parameters, except for MAE in the ‘Rustenburg’ Navel orchard, are within
the ranges stipulated by de Jager (1994) for good model performance (Figure 5.13).
© University of Pretoria
115
The MAE, which was greater than 20 % in the ‘Rustenburg’ Navel orchard, was
probably a reflection of the underestimation of transpiration by the model in autumn
and winter (Figure 5.12). The Penman-Monteith equation coupled with Jarvis canopy
conductance model (using orchard specific parameters) underestimated cumulative
transpiration by 13 % in the ‘Delta’ Valencia, 2 % in ‘Bahianinha’ and 18% in the
‘Rustenburg’ Navel orchards over the measurement period (Figure 5.14).
'Bahianinha' Navels
0.0 0.4 0.8 1.2 1.6
Pre
dic
ted tra
nspiration (
mm
)
0.0
0.4
0.8
1.2
1.6
y = 0.99x
R2 = 0.52
MAE = 17.92 %D = 0.88
RMSE = 0.04
Measured transpiration (mm)
0 1 2 3
0
1
2
3
y = 1.092x
R2
= 0.60MAE = 29.73 %
D = 0.81RMSE = 0.24
'Rustenburg' Navels
Figure 5.13. Regression analysis between predicted and measured transpiration in
the ‘Bahianinha’ and ‘Rustenburg’ Navel orchards. Statistical
parameters used include slope of regression equation, coefficient of
determination (R2), mean absolute error (MAE), Willmot coefficient of
agreement (D) and root mean square error (RMSE).
© University of Pretoria
116
'Delta' Valencias
01-Dec 01-Jan 01-Feb 01-Mar 01-Apr 01-May 01-Jun 01-Jul
0
50
100
150
200
250
Measured
Predicted
'Bahianinha' Navels
01-Dec 01-Jan 01-Feb 01-Mar 01-Apr 01-May 01-Jun 01-Jul
Cum
ula
tive tra
nspiration (
mm
)
0
20
40
60
80
100
120
140
160
180
'Rustenburg' Navels
Date
01-Sep 01-Nov 01-Jan 01-Mar 01-May 01-Jul
0
100
200
300
400
500
Figure 5.14. Cumulative measured transpiration and transpiration predicted using a
Jarvis-type model parameterised in the respective orchards for the
measurement periods. The plateau in water use graphs occurring
around April/May in the three graphs is due to periods of missing data.
© University of Pretoria
117
5.4 Discussion
The optimised hourly values of gc max obtained in the three orchards were higher than
the maximum values calculated by inverting the Penman-Monteith equation. The
value of gc max represents the ideal situation at which all functions are equal to one.
In fact the Jarvis model predicts gc max values under conditions of low VPD and high
SR and such conditions do not normally occur in the field (Whitley et al., 2009).
These values were also higher than the 6.98 to 7.96 mm s-1 that was reported by
Oguntunde et al. (2007) in sweet oranges which might mean that this parameter is
not conservative for citrus. However, the gc max obtained for the different orchards by
the optimisation process, followed in this instance, contains artefacts of the gc data
used as well as the parameters of the response functions that were attained.
The parameter kD1 is much lower than the values of 0.116 and -0.229 kPa reported
in poplar trees by Zhang et al. (1997), who used a similar form of the VPD function,
most probably indicating the differences in response of the two species to VPD. The
optimised value for the parameter kT, for optimum temperature, obtained in all three
orchards was slightly lower than 25.5 oC that was established by Oguntunde et al.
(2007). Despite the differences in canopy sizes and different environmental
conditions experienced in the orchards the response functions obtained in three
orchards in this study were very similar.
When the parameters of the Jarvis model were used to predict conductances,
derived from the sap flow data, the model was only able to account for a small
proportion (56 to 60 %) of the variance in gc. The modelling approach first suggested
by Jarvis (1976) has been widely used to predict conductances in fruit trees,
including citrus. The level at which these weather variables were able to explain the
variability in canopy conductance is slightly lower than the 83 % reported by
Oguntunde et al. (2007) in sweet oranges under semi-arid Tropical Monsoon climate,
but higher than 11 % obtained by Cohen and Cohen (1983) after fitting a similar
Jarvis-type model to measured leaf conductance of Shamouti orange trees under
Mediterranean conditions.
© University of Pretoria
118
The Penman-Monteith equation coupled with the Jarvis model (with one set of
parameters) was able to predict transpiration successfully in the three orchards over
a relatively long period. This is despite the limitations of Jarvis-type models on days
when the crop canopy is wet, or in cloudy or overcast conditions, that most studies
such as Oguntunde et al. 2007 tend to eliminate from the data set when testing the
approach. A similar modelling approach by Whitely et al. (2009) gave fairly
reasonable results in forest species over a period of 109 days. This modelling
approach may be applicable to semi-arid sub-tropical conditions where wetting of the
canopy by rainfall and occurrence of cloudy/overcast conditions are both infrequent.
The parameters of the Jarvis-type model obtained in the ‘Delta’ Valencia orchard did
not predict transpiration accurately in either the ‘Bahianinha’ or ‘Rustenburg’ Navel
orchards even though they were similar. The multiplicative form of the Jarvis model
means that the differences observed in the parameters are multiplied, hence the
large differences between measured and predicted values when using such an
approach. The fact that transpiration was overestimated in the ‘Bahianinha’ Navel
orchard and underestimated in the ‘Rustenburg’ Navel orchard, although with a
reasonable correlation, might mean that transferring such an approach amongst
orchards is not as straight-forward as proposed previously by Whitehead (1998). For
instance Villalobos et al. (2013) argued that LAI may not be the best parameter to
use when up-scaling transpiration in fruit tree orchards since it is a reflection of the
available, rather than the conducting leaf area. In order for this approach to work a
methodology for determining the leaf area active in transpiration would need to be
formulated.
There are a few possible reasons for the failure of the Jarvis-type parameters
obtained in the ‘Delta’ Valencia to accurately predict transpiration in both the
‘Bahianinha’ and ‘Rustenburg’ Navel orchards. There is evidence that stem hydraulic
conductance varies among rootstock/scion combinations in citrus (Rodríguez-Gamir
et al., 2010), which would have a direct bearing on the variations of canopy
conductance values calculated using the Penman-Monteith equation based on sap
flow measurements. Cultivar effects are probably partly reflected by the fact that
largest difference observed between the kD2 parameters of the VPD response
function was between the ‘Rustenburg’ Navel and ‘Delta’ Valencia trees. The
© University of Pretoria
119
thickness and composition of epicuticular waxes found on citrus that help to
suppress cuticular transpiration vary amongst citrus species, such that the sensitivity
of the cultivars to environmental conditions is also expected to differ (Baker et al.,
1975). Although the findings of Baker et al. (1975) were at species level, the
occurrence of such difference in cultivars and their effect on the regulation of
transpiration (and hence conductance) would be worthy of investigation.
It is also important to note that the response functions of Jarvis-type model depend
upon the physiological condition of the plants, the previous and current season in
which measurements were conducted (Jarvis, 1976). The differences in leaf
resistance brought about by the differences in crop load of these orchards that were
discussed in Chapter 4 would also be expected to manifest at canopy level. The
highest VPD recorded at the Citrusdal site is almost twice that observed at
Groblersdal during the measurement period. Citrus transpiration on which the gc
values were based has been shown to be sensitive to VPD (Oguntunde et al., 2007)
which may mean that the response of the trees under these conditions may have
differed. In addition the maximum solar radiation recorded at the Groblersdal site
was above 1000 W m-2 in both seasons as compared to a maximum of
approximately 800 W m-2 recorded at the Citrusdal site. The ability of citrus trees to
acclimatize to different levels of solar radiation, as reported by Syvertsen (1984),
may also have an effect on the responses of the citrus leaves. Since measurements
of transpiration were conducted for only one season in each orchard, the effects of
conditions in the previous seasons could not be evaluated.
5.5 Conclusion
The use of the Penman-Monteith equation coupled with a Jarvis-type model for
predicting transpiration in citrus orchards was tested. The Jarvis model had response
functions of solar radiation, temperature and vapour pressure deficit. The model was
able to predict long-term variations of citrus transpiration reasonably well only when
parameterized using data from that specific orchard. However, transferability of the
optimised parameters from one orchard to an orchard in a similar or different region
was not possible.
© University of Pretoria
120
Chapter 6: General conclusions and recommendations
6.1 Overview of study
Citrus is a one of the most important fruit crops grown across the world, as well as in
South Africa. Production of this evergreen crop in semi-arid countries, which receive
seasonal rainfall, is made possible through irrigation. The advent of micro-irrigation
systems, such as drip, provides an opportunity to match crop water requirements
and irrigation amounts, but needs to be coupled with appropriate water management
techniques (Jones, 2004). Insufficient knowledge of fruit tree water use, under these
relatively new systems, has been identified as a major factor hindering the
application and further development of existing irrigation water management tools in
South Africa (Pavel et al., 2003; Volschenk et al., 2003). This is probably because
measurements of crop water use are time consuming, expensive and require
specialized expertise, which are not always available. The situation is made worse in
orchards, where tree size is variable between different orchards and by the long time
needed to obtain good estimates (Hoffman et al., 1982; Castel et al., 1987). Water
use models have proved very useful in extrapolating measured data and predicting
water use of different crops.
The aims of this study were two-fold: i) to measure long-term transpiration of well
managed micro-irrigated citrus orchards and; ii) to test physically–based models that
can be used to predict transpiration across different citrus growing regions. This
study concentrated on the transpiration component since it forms the larger portion
of orchard water use in micro-irrigated fruit tree orchards (Reinders et al., 2010) and
is directly related to productivity (Villalobos et al., 2013). In order to meet the
objectives, sap flow measurements were performed using the heat ratio method
(HRM) as described by Burgess et al. (2001) in two orchards in South Africa. The
two orchards planted with ‘Delta’ Valencia and ‘Bahianinha’ Navel orange trees were
located in the summer rainfall area. A data set from an orchard planted with
‘Rustenburg’ Navel orange trees that was located in the winter rainfall area was
sourced for model validation. Weather variables, which include solar radiation,
temperature, relative humidity and wind speed, were measured using automatic
weather stations close to the orchards for modelling purposes.
© University of Pretoria
121
6.2 Calibration of the heat ratio method (HRM) sap flow technique for long-term
transpiration measurement in citrus orchards
The HRM was calibrated against eddy covariance measurements in winter (31July to
3 August 2008) and autumn (21 to 25 May 2009) in the ‘Delta’ Valencia orchard,
when soil evaporation was considered negligible (Chapter 2). Calibration was done
by adjusting the wounding width to ensure that average transpiration of sample trees
matched crop evapotranspiration (ETc) measured above the orchard using the eddy
covariance method. The virtual wound widths that matched mean transpiration of
sample trees to ETc for winter and autumn periods were 3.3 and 3.1 mm,
respectively. The similarity of the virtual wound widths obtained during the two
measurement periods (winter and autumn) and the subsequent successful use of an
average value (3.2 mm) suggests that wounding does not increase with time and a
single field calibration of the HRM using eddy covariance measurements is sufficient
for measuring transpiration in citrus orchards. The requirement for a wound width
greater than the observed 2 mm width in citrus was necessitated by the fact that
citrus had non-homogeneous functional wood as determined by the anatomical
assessments. Non-homogeneity of wood means that the ideal heat pulse theory will
not apply to this wood and a correction factor is therefore required. This is in
agreement with previous studies in kiwifruit (Green and Clothier, 1988) and citrus
(Fernández et al., 2006). In addition citrus wood reacted to drilling and heating by
deposition of phenols and gums that caused xylem blockages which extended
beyond what could be seen by the naked eye, as revealed by the anatomical
assessments in this study. Once the calibration of the HRM was performed, the
method was then used to quantify transpiration in the three orchards.
6.3 Measurement of long-term transpiration and transpiration coefficients in
citrus orchards
In order to address the problem of lack of information of water use in citrus orchards,
which was previously highlighted (Pavel et al., 2003; Volschenk et al., 2003), the
HRM was employed to quantify long-term transpiration rates in three citrus orchards
(Chapter 3). Average daily transpiration in the three orchards ranged from 0.77 mm
day-1 to 2.26 mm day-1 depending on canopy size, season and climate (summer
versus winter rainfall areas). Total seasonal transpiration for the two orchards in
© University of Pretoria
122
which measurements covered an entire production season, i.e. ‘Delta’ Valencia and
‘Rustenburg’ Navel orchards were 682 mm and 672 mm, respectively. This study
confirmed the variations of transpiration in citrus orchards as previously reported
(García Petillo and Castel, 2007). Whilst the measurement of transpiration in the
three orchards is not exhaustive of all orchard conditions in South Africa, it does
provide a basis for the understanding of the drivers of transpiration. Sap flow-based
measurements of transpiration obtained in this study presented a valuable data set
for testing models of citrus transpiration that can be used to extrapolate measured
transpiration to other growing regions of the world.
Transpiration coefficients (Kt) determined in the three orchards ranged between 0.28
and 0.71. The Kt values were almost constant throughout the seasons in the summer
rainfall area, and significantly higher in the winter months than in the summer months
in the winter rainfall area. This indicated that transpiration does not increase at the
same rate as reference evapotranspiration (ETo) during the summer months and
confirmed the existence of strong regulation mechanisms within the citrus tree
hydraulic path that was previously reported (van Bavel et al. 1967; Kriedemann and
Barrs 1981; Sinclair and Allen 1982). It is also indicative of fact that citrus trees are
only able to supply a certain maximum amount of water to the canopy at high ETo (≥
5 mm day-1), even when soil water is not limiting, as suggested by Sinclair and Allen
(1982).
Differences in measured Kt values between the orchards in Groblersdal, despite
similar ETo in both years, suggests an effect of canopy size on transpiration. On the
other hand differences in the trends of Kt values between the two orchards located in
Groblersdal and the one in Citrusdal in winter could be linked to differences in VPD.
Vapour pressure deficit was reported as the dominant regulator of transpiration of
citrus by Kriedemann and Barrs (1981) and Oguntunde et al. (2007) when soil water
is not limiting. The clear differences between Kt values in the three orchards and
those published in literature (including the FAO56) indicate that the Kt values are
orchard specific. This result confirmed the variation of Kt values of citrus orchards
reported previously (Rana et al. 2005; Villalobos et al., 2009; Marin and Angelocci,
2011; Villalobos et al., 2013). Crop coefficients determined in one orchard can
therefore not be directly applied or transferred to different growing regions of the
© University of Pretoria
123
world. This emphasizes the need to develop an easy method for estimating site-
specific crop coefficients, in order to improve water management in citrus orchards.
6.4 Estimation of transpiration coefficients from plant height and canopy cover
The significance of the role of the resistance to water transport within the citrus tree
transpiration stream was highlighted during the development Kt values based on
crop height and canopy cover following the method suggested by Allen and Pereira
(2009) (Chapter 4). Good agreements between measured and estimated Kt values
were only obtained after back-calculating leaf resistance (rl) using measured
transpiration values, as recommended by Allen and Pereira (2009). This study
established that the leaf resistance of citrus may be higher than what was suggested
by Allen and Pereira (2009) for a generic citrus orchard. The rl values of ranging from
419 to 2 694 s m−1 obtained through this procedure were comparable to those
ranging from 200 to 8 280 s m−1 reported in literature (Cohen and Cohen 1983;
Dzikiti et al. 2007; Pérez-Pérez et al., 2008). It was also established that contrary to
the constant rl value (420 s m-1) suggested by Allen and Pereira (2009), leaf
resistance increased for all orchards during the summer period. Allen and Pereira
(2009) recommended that improvements in estimating the rl term are needed to
improve the accuracy of the methodology. In that regard a relationship between
mean monthly rl and ETo in the ‘Rustenburg’ Navel orchard provided a means of
estimating mean leaf resistance, which estimated Kt values with a reasonable degree
of accuracy in the three orchards. However, this relationship only provided good
seasonal estimates of transpiration, which means that it can only be useful for
irrigation planning. A more mechanistic model capable of capturing the dynamics of
the canopy conductance in citrus varieties is required to predict transpiration of citrus
trees on day-to-day basis for purposes such as irrigation scheduling.
6.5 Predicting transpiration of citrus orchards using the Penman-Monteith
equation coupled with a Jarvis-type canopy conductance model
The use of a more mechanistic approach that incorporates a dynamic canopy
conductance term was also tested (Chapter 5). This was in the form of the Penman-
Monteith equation, coupled with the estimation of canopy conductance using a
Jarvis-type model, as proposed by Oguntunde et al. (2007). It was demonstrated that
© University of Pretoria
124
once optimised a single set of response functions for canopy conductance, obtained
in each orchard, can be used to predict transpiration accurately in the respective
orchard for periods over six months. This approach may be applicable to sub-tropical
conditions where wetting of the canopy by rainfall and occurrence of cloudy/overcast
conditions are both infrequent. This result is similar to what was established by
Whitely et al. (2009) who also reported that this approach could be applied in a forest
species for a fairly long time. The ability of a single set of response functions to
predict transpiration when the Jarvis-type model was parameterized using data in the
respective orchards does put hope in the method as a way of mechanistically
predicting citrus transpiration. It was also shown that the parameters of the Jarvis
model obtained in an orchard in the summer rainfall are not suitable for predicting
conductance (and hence transpiration) in the same or different region. This was
attributed to the fact that the parameters obtained in one orchard contain artifacts of
the measured data that may not exist in the other orchards. The factors that could
potentially influence these parameters and hinder the transferability of the Jarvis-
type model, which could not be ascertained in this study, include differences in
varieties, climatic conditions and crop load. This approach will only be applicable if
parameterized with measured data.
6.6 Recommendations for future research
Knowledge of water use of citrus orchards is crucial for irrigation planning and
management. In this study transpiration of three citrus orchards was measured, and;
the data was used to test some of the models that can be used to estimate water use
in these orchards. Nevertheless, questions related to water use of citrus orchards
still remain unanswered. Measurements of transpiration in three orchards presented
in this study are not exhaustive on how much water use is used by citrus under all
environmental conditions and management strategies in South Africa. Measurement
of transpiration should be conducted in different growing regions under the different
strategies to consolidate the base information of citrus orchard transpiration attained
in this study. Such monitoring experiments should aim to address water use of the
trees in relation to factors possibly impacting transpiration e.g. canopy size,
rootstock/scion combinations and crop load. Citrus water use measurements should
be conducted simultaneously with environmental monitoring to allow for further
© University of Pretoria
125
improvement of the modelling approaches that were evaluated in this study. There is
room for improvement in the estimation of the rl term so as to use the method
predicting transpiration on daily time scale. The approach for transferring the Jarvis-
type model amongst orchards needs further research. A starting point would be to
determine a maximum canopy conductance value of citrus that is independent of the
measured transpiration data for parameterization of such a model. Information on
water use of specific rootstock/scion combinations obtained for long periods covering
consecutive seasons can also be used to answer question pertaining to the
transferability of Jarvis-type canopy conductance model amongst different varieties
that arose in this study. The exploration of other modelling approaches such as a
combination of a crop coefficient and a maximum transpiration rate as envisaged by
Sinclair and Allen (1982) should be considered as well. Using such a model, it will be
important to be able to determine what the maximum transpiration rate would be for
a certain sized canopy, which may be addressed by way of extensive measurements
mentioned above.
© University of Pretoria
126
References
ALLEN, R., 2008. Quality assessment of weather data and micrometeorological flux
– impacts on evapotranspiration calculation. J. Agric. Meteorol. 64, 191-204.
ALLEN, R.G., SMITH, M., PERRIER, A., AND PEREIRA, L.S., 1994. An update for
the definition of reference evapotranspiration. ICID Bulletin 43, 1-34.
ALLEN, R.G., PEREIRA, L.S., RAES, D., SMITH, M., 1998. Crop evapotranspiration:
guidelines for computing crop water requirements, Irrigation and Drainage
Paper 56. United Nations FAO, Rome
http://www.fao.org/docrep/X0490E/X0490E00.htm
ALLEN, R.G., PEREIRA, L.S., SMITH, M., RAES, D., WRIGHT, J.L., 2005. FAO-56
dual crop coefficient method for estimating evaporation from soil and
application extensions. J. Irrig. Drain. Eng. ASCE 131, 2–13.
ALLEN, R.G., PEREIRA, L.S., 2009. Estimating crop coefficients from fraction of
ground cover and height. Irrig. Sci. 28, 17-34.
ALLEN, R.G., PRUITT W.O., WRIGHT, J.L., HOWELL, T.A., VENTURA, F.,
SNYDER, R., ITENFISU, D., STEDUTO, P., BERENGENA, J., BASELGA
YRISARRY, J., SMITH, M., PEREIRA, L.S., RAES, D., PERRIER, A.,
ALVES, I., WALTER, I., ELLIOT, R., 2006. A recommendation on
standardised surface resistance for hourly calculation of reference ETo by
the FAO56 Penman-Monteith method. Agric. Water Manage. 81, 1-22.
ALLEN, R.G., WRIGHT, J.L., PRUITT, W.O., PEREIRA, L.S., 2007. Water
requirements. In: Design and operation of farm irrigation systems 2nd edn.
American Society of Agricultural Engineers, pp 208-288.
ALLEN, R.G., KJAERSGAARD, J., GARCÍA, M., TASUMI, M., TREZZA, R., 2008.
Fine-tuning components of inverse-calibrated, thermal-based remote
sensing models for evapotranspiration. In: Proc. 17th Pecora Conference on
the Future of Land Imaging , Denver, November 2008.
© University of Pretoria
127
ALLEN, R.G., PEREIRA, L.S., HOWELL, T.A., JENSEN, M.E., 2011.
Evapotranspiration information reporting: I. Factors governing measurement
accuracy. Agric. Water Manage. 98, 899-920.
ALLEN, S.J., GRIME, V.L., 1995. Measurements of transpiration from savannah
shrubs using sap flow gauges. Agric. For. Meteorol. 75, 23-41.
ANNANDALE, J., STOCKLE, C., 1994. Fluctuation of crop evapotranspiration
coefficients with weather: a sensitivity analysis. Irrig. Sci 15, 1–7.
ANNANDALE, J.G., JOVANOVIC, N.Z., CAMPBELL, G.S., SAUTOY, N.D.,
BENADE, N., 2003. A two-dimensional water balance model for micro-
irrigated hedgerow tree crops. Irrig. Sci. 22,157-170.
ASCE-EWRI, 2005. The ASCE Standardized Reference Evapotranspiration
Equation. Report 0-7844-0805-X, ASCE Task Committee on Standardization
of Reference Evapotranspiration. Reston, VA., American Soc. Civil
Engineers.
AYARS, J.E., JOHNSON, R.S., PHENE, C.J., TROUT, T.J., CLARK, D.A. & MEAD,
R.M., 2003. Water use by drip-irrigated late-season peaches. Irrig. Sci. 22,
187-194.
BALDOCCHI D. D., LUXMOORE, R. J., HATFIELD, J. L., 1991. Discerning the
forest from the trees: an essay on scaling canopy stomatal conductance.
Agric. For. Meteorol. 54, 197-226.
BARRETT, D.J., HATTON, T.J., ASH J.E., BALL. M.C., 1995. Evaluation of the heat
pulse velocity technique for measurement of sap flow in rainforest and
eucalypt forest species of south-eastern Australia. Plant Cell Environ. 18,
463–469.
BECKER, P., 1998. Limitations of a compensation heat pulse velocity system at low
sap flow: implications for measurements at night and in shaded trees. Tree
Physiol. 18, 177-184.
© University of Pretoria
128
BENYON, R.G., 1999. Nighttime water use in an irrigated Eucalyptus grandis
plantation. Tree Physiol. 19, 853–859.
BIJZET Z, 2006. Climatic Requirements. In: De Villiers, E.A., Jourbet, P.H., (Eds.)
The cultivation of citrus, Arc-Institute For Tropical And Subtropical Crops. pp
28-42.
BLEBY, T.M., BURGESS, S.S.O., ADAMS, M.A., 2004. A validation, comparison
and error analysis of two heat-pulse methods for measuring sap flow in
Eucalyptus marginata saplings. Functional Plant Biology 31, 645-658.
BOOTE KJ, JONES JW and PICKERING, NG. 1996. Potential uses and limitations
of crop models. Agron. J. 88: 704-716.
BURBA, G., ANDERSON D., 2008. Introduction to the eddy covariance method:
general guidelines and conventional workflow. LiCor Biosciences, LiCor
Corporation, Lincoln, NE.
BURGESS, S. S.O., ADAMS, M. A., TURNER N. C., ONG C. K., 1998. The
redistribution of soil water by tree root systems. Oecologia 115, 306-311.
BURGESS, S.S.O., PATE, J.S., ADAMS M.A., DAWSON, T.E., 2000. Seasonal
water acquisition and redistribution in the Australian woody phreatophyte,
Banksia prionotes. Ann. Bot. 85, 215–224.
BURGESS, S.S.O., ADAMS, M.A., TURNER, N.C., BEVERLY, C.R., ONG, C.K.,
KHAN, A.A.H., BLEBY, T.M., 2001. An improved heat pulse method to
measure low and reverse rates of sap flow in woody plants. Tree Physiol. 21,
589–598.
CAMPBELL, G.S., NORMAN J.M., 1998. An introduction to environmental
biophysics. Springer-Verlag New York, Inc.
CARR MK, 2012. The water relations and irrigation requirements of citrus (Citrus
spp.): A review. Expl. Agric. 48, 347-377
© University of Pretoria
129
CASPARI, H.W., GREEN, S.R., EDWARDS, W.R.N., 1993. Transpiration of well-
watered and water-stressed Asian pear trees as determined by lysimetry,
heat-pulse, and estimated by a Penman-Monteith model. Agric. For.
Meteorol. 67, 13-27.
CASTEL, J.R., 1997. Evapotranspiration of a drip-irrigated clementine citrus tree in a
weighing lysimeter. Acta Hort. 449, 91-98.
CASTEL, J.R., BAUTISTA I., RAMOS C., CRUZ G., 1987. Evapotranspiration and
irrigation efficiency of mature orange orchards in Valencia (Spain). Irr. Drain.
Sys. 3, 205-217.
CASTEL, J.R., 2000. Water use of developing citrus canopies in Valencia, Spain. In
Proceedings of the IX International Citrus Congress, Orlando, Florida, pp
223-226.
CITRUS GROWERS ASSOCIATION OF SOUTHERN AFRICA, 2015a. Annual
report 2015. www.cga.co.za (accessed on 20 March 2015)
CITRUS GROWERS ASSOCIATION OF SOUTHERN AFRICA, 2015b. Key industry
statistics 2015.
CONCEIÇÃO, N., FERREIRA, M.I., 2008. Combination of sap flow and eddy
covariance techniques to obtain long term transpiration in a pear orchard.
Acta Hort. 846, 53-60.
COHEN, S., COHEN, Y., 1983. Field studies of leaf conductance response to
environmental variables in citrus. J. Appl. Ecol. 20, 561–570.
COHEN, Y., COHEN, S., CANTUARIAS-AVILES, T., SCHILLER, G., 2008.
Variations in the radial gradient of sap velocity in trunks of forest and fruit
trees. Plant Soil 305, 49-59.
CONSOLI, S., PAPA, R., 2013. Corrected surface energy to measure and model the
evapotranspiration of irrigated orange orchards in semi-arid Mediterranean
conditions. Irrig. Sci. 31, 1159-1171.
© University of Pretoria
130
CONSOLI, S., O’CONNELL, N., AND SNYDER, R. 2006. Estimation of
evapotranspiration of different-sized navel-orange tree orchards using
energy balance. J. Irrig. Drain Eng. 132, 2–8.
DAAMEN, C.C., DUGAS, W., PRENDERGAST, P., JUDD, M., MCNAUGHTON, K.,
1999. Energy flux measurements in a sheltered lemon orchard. Agric. For.
Meteorol. 93, 171-183.
DAMOUR, G., SIMONNEAU, T., COCHARD, H., URBAN, L., 2010. An overview of
models of stomatal conductance at the leaf level. Plant, Cell Environ. 33,
1419-1438.
DE JAGER, J., 1994. Accuracy of vegetation evaporation ratio formulae for
estimating final wheat yield. Water SA 20, 307-314
DRAGONI, D., LAKSO, A.N., PICCIONI, R.M., 2005. Transpiration of apple trees in
a humid climate using heat pulse sap flow gauges calibrated with whole-
canopy gas exchange chambers. Agric. For. Meteorol. 130, 85-94.
DOORENBOS, J., PRUITT, W.O., 1977. Crop water requirements. Irrigation And
Drainage Paper No. 24. Fao, Rome, Italy.
DUAN, W., FAN, P.G., WANG, L.J., LI, W.D., YAN, S.T., LI, S.H., 2008.
Photosynthetic response to low sink demand after fruit removal in relation to
photoinhibition and photoprotection in peach trees. Tree Physiol. 28, 123–
132
DUGAS, W.A., MAYEUX JR, H.S., 1991. Evaporation from rangeland with and
without honey mesquite. J. Range Manage. 161-170.
DU PLESSIS, H.M., 1985. Evapotranspiration of citrus as affected by soil water
deficit and soil salinity. Irrig. Sci. 6, 51-61.
DWAF (DEPARTMENT OF WATER AFFAIRS AND FORESTRY, SOUTH AFRICA)
(2013) National Water Resource Strategy. (2nd Edition.). June 2013.
Department of Water Affairs and Forestry,Pretoria, South Africa.
© University of Pretoria
131
DZIKITI, S., STEPPE, K., LEMEUR, R., MILFORD, J.R., 2007. Whole-tree level
water balance and its implications on stomatal oscillations in orange trees
[Citrus Sinensis (L.) Osbeck] under natural climatic conditions. J. Exp. Bot.
58, 1893–1901.
DZIKITI, S., SCHACHTSCHNEIDER, K., NAIKEN, V., GUSH M., MOSES, G., LE
MAITRE, D.C., 2013. Water relations and the effects of clearing invasive
Prosopis trees on ground water in an arid environment in the Northern Cape,
South Africa. J. Arid Environ. 90, 103-113.
DZIKITI, S., VERREYNNE, S.J., STUCKENS, J., STREVER, A., VERSTRAETEND,
W.W., SWENNEN, R., THERONA, K.I., COPPIN, P., 2011. Seasonal
variation in canopy reflectance and its application to determine the water
status and water use by citrus trees in the Western Cape, South Africa.
Agric. For. Meteorol. 151, 1035–1044
EDWARDS, W.R.N., BECKER, P., ÈERMÁK, J., 1997. A unified nomenclature for
sap flow measurements. Tree Physiol. 17, 65-67.
EL NADI, AH, HUDSON, JP, 1965. Effects of crop height on evaporation from
lucerne and wheat grown in lysimeters under advective conditions in the
Sudan. Exp. Agric. 1, 289-298.
ER-RAKI, S., CHEHBOUNI A., BOULET G., WILLIAMS D.G., 2010. Using the dual
approach of FAO-56 for partitioning ET into soil and plant components for
olive orchards in a semi-arid region. Agric. Water Manage. 97, 1769-1778.
FAO-AQUASTAT, 2003. Information system on water and agriculture: review of
world water resources by country. Food and Agricultural Organisation of the
World, Rome. http://www.fao.org (accessed on 16 May 2011).
FAO, 2012. FAO statistics. Website: http:/apps.fao.org/default.htm (accessed
December 2013).
© University of Pretoria
132
FARES, A., DOGAN, A., ABBAS, F., PARSONS, L.R., OBREZA, T.A., MORGAN,
K.T., 2008. Water balance components in a mature citrus orchard. Soil Sci.
Soc. Am. J. 72, 578-585.
FARES, A., ALVA, A. K. 1999. Estimation of citrus evapotranspiration by soil water
mass balance. Soil Sci. 164, 302-310.
FERNÁNDEZ, J.E., DURAN, P.J., PALOMO, M.J., DIAZ-ESPEJO, A., CHAMORRO
V., GIRÓN, I.F., 2006. Calibration of sap flow estimated by the
compensation heat pulse method in olive, plum and orange trees:
Relationship with xylem anatomy. Tree Physiol. 26, 719-728.
FERNÁNDEZ, J.E., PALOMO, M.J., DIAZ-ESPEJO, A., CLOTHIER, B.E., GREEN,
S.R., GIRON, I.F., MORENO, F., 2001. Heat-pulse measurements of sap
flow in olives for automating irrigation: tests, root flow and diagnostics of
water stress. Agric. Water Manage. 51, 99-123.
FOKEN, T, LEUNING R, ONLEY SR, MAUDER M, AUBINET M. 2012. Corrections
and data quality control. In Aubient M, Vesala T, Papale D. (eds). Eddy
Covariance: A Practice Guide to Measurement and Data Analysis. Springer,
New York. p 85-131.
GARCÍA PETILLO, M., CASTEL, J.R., 2007. Water balance and crop coefficient
estimation of a citrus orchard in Uruguay. Spanish J. of Agric. Res. 5, 232-
243.
GIORIO, P., GIORIO, G., 2003. Sap flow of several olive trees estimated with the
heat-pulse technique by continuous monitoring of a single gauge. Environ.
Exp. Bot. 49, 9-20.
GIRONA, J., MATA, M., FERERES, E., GOLDHAMER, D.A., COHEN, M., 2002.
Evapotranspiration and soil water dynamics of peach trees under water
deficits. Agric. Water Manage. 54, 107-122.
GIRONA, J., MARSAL, J., MATA, M., DEL CAMPO, J., 2003. Pear crop coefficients
obtained in a large weighing lysimeter. Acta Hort. 664, 277-281.
© University of Pretoria
133
GLIME, J. M., AND WAGNER, D. H., 2013. Laboratory Techniques: Slide
Preparation and Stains. In: Glime, J. M. (Ed) Bryophyte 2-2-1 Ecology.
Volume 3. Methods. E-book sponsored by Michigan Technological University
and the International Association of Bryologists.
GOODWIN, I., WHITFIELD, D.M., CONNOR, D.J., 2006. Effects of tree size on
water use of peach (Prunus Persica L. Batsch). Irrig. Sci. 24, 59-68.
GUSH, M.B., 2008. Measurement of water-use by Jatropha curcas L. using the heat-
pulse velocity technique. Water SA 34, 1-5.
GRANIER, A., LOUSTAU, D., 1994. Measuring and modelling the transpiration of a
maritime pine canopy from sap-flow data. Agric. For. Meteorol. 71, 61-81.
GREEN, S.R.,CLOTHIER, B.E., 1988. Water use of kiwifruit vines and apple trees by
the heat-pulse technique. J. Exp. Bot. 39, 115-123.
GREEN, S., CLOTHIER, B., JARDINE, B., 2003. Theory and practical application of
heat pulse to measure sap flow. Agron. J. 95, 1371-1379.
GREEN, S.R., MCNAUGHTON, K.G., 1997. Modelling effective stomatal resistance
for calculating transpiration from an apple tree. Agric. For. Meteorol. 83, 1-
26.
GREEN G.C., MORESHET S., 1979. An analysis of seasonal water use
characteristics of Valencia orange trees in the Sundays River Valley. Crop
Prod. 8, 179-183.
HATTON, T.J., CATCHPOLE, E.A., VERTESSY, R.A., 1990. Integration of sapflow
velocity to estimate plant water use. Tree Physiol. 6, 201-209.
HATTON, T.J., MOORE, S.J., REECE, P.H., 1995. Estimating stand transpiration in
a Eucalyptus populnea woodland with the heat pulse method: measurement
errors and sampling strategies. Tree Physiol. 15, 219 225.
© University of Pretoria
134
HOFFMAN, G.J., OSTER, J.D., ALVES, W.J., 1982. Evapotranspiration of mature
orange trees under drip irrigation in an arid climate. Trans. Am. Soc. Agric.
Eng. 25, 992–996.
HULTINE, K.R., NAGLER, P.L., MORINO, K., BUSH, S.E., BURTCH, K.G.,
DENNISON, P.E., GLENN, E.P., EHLERINGER, J.R., 2010. Sap flux-scaled
transpiration by tamarisk (Tamarix spp.) before, during and after episodic
defoliation by the saltcedar leaf beetle (Diorhabda carinulata). Agric. For.
Meteorol. 150, 1467–1475.
IGLESIAS, D.J., LLISO, I., TADEO, F.R., TALON, M., 2002. Regulation of
photosynthesis through source: sink imbalance in citrus is mediated by
carbohydrate content in leaves. Physiol Plant. 116, 563–572
IPCC, 2007. Fourth assessment report of Intergovernmental Panel on Climate
Change. Cambridge University Press, Cambridge, UK.
JARA, J., STOCKLE, C.O., KJELGAARD, J., 1998. Measurement of
evapotranspiration and its components in a corn (Zea Mays L.) field. Agric.
For. Meteorol. 92,131-145.
JARVIS, P.G., 1976. The interpretation of the variations in leaf water potential and
stomatal conductance found in canopies in the field. Phil. Trans. R. Soc.
Lond. 273, 593-610.
JONES H. G., 1985. Physiological control of water status in temperate and
subtropical fruit trees. Hort. Rev. 7, 301-345.
JONES H. G., HAMER P. J. C., HIGGS K. H., 1988. Evaluation of various heat-pulse
methods for estimation of sap flow in orchard trees: comparison with
micrometeorological estimates of evaporation. Trees 2, 250-260
JONES, H.G., 2004. Irrigation Scheduling: Advantages and pitfalls of plant-based
methods. J. Exp. Bot. 55, 2427-2436.
© University of Pretoria
135
KALMA J.D., STANHILL G., 1969. Transpiration, evaporation and deep drainage
losses from an orange plantation. Israel J. Agric. Res. 19, 11-24.
KAUFMANN, M., 1977. Citrus—a case study of environmental effects on plant water
relations. Proc. Int. Soc. Citric (Orlando) 1, 52–57.
KHAIRI, M.M., HALL, A.E., 1976. Temperature and humidity effects on net
photosynthesis and transpiration of citrus. Physiol. Plant. 36, 29-34.
KOO, R.C.J., SITES, J.W., 1955. Results of research and response of citrus to
supplemental irrigation. Soil Sci. Soc. Fla Proc. 15, 180-190.
KRIEDEMANN, P., 1971. Photosynthesis and transpiration and function of gaseous
diffusive resistances in orange leaves. Physiol. Plant. 24, 218-225.
KRIEDEMANN, P., BARRS, H., 1981. Citrus orchards. In: Kozlowski TT (ed) Water
Deficits and Plant Growth, Academic Press, New York, pp 325-417.
LANDSBERG, J.J., POWELL, D.B.B., 1973. Surface exchange characteristics of
leaves subject to mutual interference. Agric. Meteorol. 12, 169-184.
LEVY, Y., KAUFMANN, M.R., 1976. Cycling of leaf conductance in citrus exposed to
natural and controlled environments. Can. J. Bot. 54, 2215-2218.
MARIN, F.R., ANGELOCCI, L.R., 2011. Irrigation requirements and transpiration
coupling to the atmosphere of a citrus orchard in Southern Brazil. Agric.
Water Manage.98, 1091-1096.
MARSHALL, D.C., 1958. Measurement of sap flow in conifers by heat transport.
Plant Physiol. 33, 385-396.
MAUDER, M., FOKEN, T., CLEMENT, R., ELBERS, J., EUGSTER, W.,
GR¨UNWALD, T., HEUSINKVELD, B., KOLLE, O., 2008. Quality control of
CarboEurope flux data - part 2: inter-comparison of eddy-covariance
software. Biogeosciences 5, 451–462.
© University of Pretoria
136
MCNAUGHTON, K.G., JARVIS, P.G., 1983. Predicting effects of vegetation changes
on transpiration and evaporation. In: Koslowsky, T.T., (Ed.) ‘Water deficit
and plant growth’. Academic Press, New York, pp. 1–47.
MEYER, W.S., GREEN, G. C., 1981. Comparison of stomatal action of orange,
soybean and wheat under field conditions. Aust. J. Plant Physiol. 8, 65-76.
MEYERS, T.P., BALDOCCHI, D.P., 2005. Current micrometerological flux
methodologies with applications in agriculture. In: Hatfield, J.L., Baker, J.M.
(Eds.) Micrometeorology in Agricultural Systems. American Society of
Agronomy, pp. 381–396.
MONTEITH, J. L., UNSWORTH. M.H., 1990. Principles of environmental physics.
Edward and Arnold, UK.
MORESHET, S., GREEN, G.C., 1984. Seasonal trends in hydraulic conductance of
field-grown ‘Valencia' orange trees. Sci. Hort. 23,169-180.
NAOR, A., NASCHITZ, S., PERES, M., GAL, Y., 2008. Responses of apple fruit size
to tree water status and crop load. Tree Physiol. 28, 1255-1261.
NEBAUER, S.G., ARENAS, C., RODRÍGUEZ-GAMIR, J., BORDÓN, Y.,
FORTUNATO-ALMEIDA, A,, MONERRI, C., GUARDIOLA, J.L., MOLINA,
R.V., 2013. Crop load does not increase the photosynthetic rate in Citrus
leavesunder regular cropping conditions. A study throughout the year. Sci.
Hortic. 160, 358–365
NORTES, P.A., BAILLE, A., EGEA, G., GONZÁLEZ-REAL, M.M., DOMINGO, R.,
2008. Comparison of almond tree transpiration determined by sap flow
measurements and lysimetry. Acta Hort. 846, 359-366.
OGUNTUNDE, P.G., VAN DE GIESEN, N., 2005. Water flux measurement and
prediction in young cashew trees using sap flow data. Hydrol. Process. 19,
3235-3248.
© University of Pretoria
137
OGUNTUNDE, P.G., VAN DE GIESEN, N., SAVENIJE, H.H.G., 2007. Measurement
and modelling of transpiration of a rain-fed citrus orchard under sub-humid
tropical conditions. Agric. Water Manage. 87, 200-208.
PAVEL, E.W., VANASSCHE, F.M.G., GROSSMAN, Y.L., 2003. Optimisation of
irrigation management in mango trees by determination of water and carbon
demands to improve water use efficiency and fruit quality. Report to the
Water Research Commission. Report no. 1136/1/03.
PAW U, K.T., QUI, J., SU, H.B., WATANABE, T., BRUNNET, Y., 1995. Surface
renewal analysis: a new method to obtain scalar fluxes. Agric. For. Meteorol.
74, 119-137.
PEREIRA, L.S., ALLEN, R. G., SMITH, M., RAES, D., 2015. Crop evapotranspiration
estimation with FAO56: Past and future. Agric. Water Manage. 147, 4-20.
PEREIRA, A.R., GREEN, S.R., NOVA, N.A.V., 2006. Penman–Monteith reference
evapotranspiration adapted to estimate irrigated tree transpiration. Agric.
Water Manage. 83, 153-161.
PÉREZ-PÉREZ, J., ROMERO, P., NAVARRO, J., BOTÍA, P., 2008. Response of
sweet orange cv ‘Lane late’ to deficit irrigation in two rootstocks.I: water
relations, leaf gas exchange and vegetative growth. Irrig. Sci. 26, 415–425
PERNICE, F., MOTISI, A., ROSSI, F., GEORGIADIS, T., NARDINO, M., FONTANA,
G., DIMINO, G.,DRAGO, A., 2009. Micrometeorological and sap flow
measurement of water vapour exchanges in olive: Scaling up from canopy to
orchard. Acta Hort. 846, 159-166.
POBLETE-ECHEVERRÍA, C., ORTEGA-FARIAS, S., ZUǸIGA, M., FUENTES, S.,
2012. Evaluation of compensated heat-pulse velocity method to determine
vine transpiration using combined measurements of eddy covariance system
and microlysimeters. Agric. Water Manage. 109, 11-19.
© University of Pretoria
138
RANA, G., KATERJI, N., DE LORENZI, F., 2005. Measurement and modelling of
evapotranspiration of irrigated citrus orchard under Mediterranean
conditions. Agric. For. Meteorol. 128, 199-209.
REINDERS, F.B., VAN DER STOEP, I., LECLER, N.L., GREAVES, K.R.,
VAHRMEIJER, J.T., BENADE, N., DU PLEISS, F.J., VAN HEERDEN, P.S.,
STEYN, J.M., GROVE, B., JUMMAN, A. & ASCOUGH, G., 2010. Standards
and guidelines for improved efficiency of irrigation water use from dam wall
release to root zone application. Report to the Water Research Commission.
Report no. TT465/10.
RIBEIRO, R. V., MACHADO C. E., 2007. Some aspects of citrus ecophysiology in
subtropical climates: re-visiting photosynthesis under natural conditions.
Braz. J. Plant Physiol., 19, 393-411.
RODRÍGUEZ-GAMIR, J., INTRIGLIOLO, D.S., PRIMO-MILLO, E., FORNER-GINER,
M., 2010. Relationships between xylem anatomy, root hydraulic conductivity,
leaf/root ratio and transpiration in citrus trees on different rootstocks. Physiol.
Plant 139, 159–169
ROGERS, J., ALLEN, L., CALVERT, D. 1983. Evapotranspiration from a humid-
region developing citrus groves with grass cover. Trans. ASAE 26, 1778–
1783
SALISBURY, F.B., ROSS, C.W., 1992. Plant Physiology. 4th Edition, Wadsworth,
Belmont, California.
SAMANI, Z., BAWAZIR, S., SKAGGS, R, LONGWORTH, J., PINON, A., TRAN, V.,
2011. A simple irrigation scheduling approach for pecans. Agric. Water
Manage. 98, 661–664
SEPASKHAH A.R., KASHEFIPOUR S.M., 1995. Evapotranspiration and crop
coefficient of sweet lime under drip irrigation. Agric. Water Manage. 27, 331-
340.
© University of Pretoria
139
SHALHEVET, J. AND LEVY, Y., 1990. Citrus trees. In: Stewart, B.A., Nielsen D. R.
(Eds.) Irrigation of agricultural crops. Wisconsin, American Society of
Agronomy. pp 951–986.
SINCLAIR, T., ALLEN L., 1982. Carbon dioxide and water vapour exchange of
leaves on field-grown citrus trees. J. Exp. Bot. 33, 1166-1175
SIPPEL A.D., 2006. Origins of citrus and its history in South Africa. In: De Villiers,
E.A., Jourbet, P.H., (Eds.) The cultivation of citrus. Arc-Institute For Tropical
And Subtropical Crops. pp 3-9.
SMITH, D.M., ALLEN, S.J., 1996. Measurement of sap flow in plant stems. J. Exp.
Bot. 47, 1833-1844.
SMAJSTRLA A.G., HARRISON D.S., TAI C., CLAPP D., 1982. Water budget of
crown flood irrigated citrus. Proc. Fla State Hort. Soc. 95, 11-14.
SNYDER, R. AND O’CONNELL, N., 2007. Crop coefficients for micro-sprinkler-
irrigated, clean-cultivated, mature citrus in an arid climate. J. Irrig. Drain.
Eng. 133, 43–52.
SOEGAARD H, BOEGH E. 1995. Estimation of evapotranspiration from a millet crop
in the Sahel combining sap flow, leaf area index and eddy correlation
technique. J. Hydrol. 166, 265-82.
SPERRY, J., HACKE, U., OREN, R., COMSTOCK, J., 2002. Water deficits and
hydraulic limits to leaf water supply. Plant Cell Environ. 25, 251–263
STEPPE, K., DE PAUW, D.J.W., DOODY, T.M., TESKEY, R.O., 2010. A comparison
of sap flux density using thermal dissipation, heat pulse velocity and heat
field deformation methods. Agric. For. Meteorol.150, 1046-1056.
SWANSON, R.H., 1994. Significant historical developments in thermal methods for
measuring sap flow in trees. Agric. For. Meteorol. 72, 113-132.
SWANSON, R.H.,WHITFIELD, D.W.A., 1981. A Numerical Analysis Of Heat Pulse
Velocity Theory and Practice. J. Exp. Bot. 32, 221-239.
© University of Pretoria
140
SWINBANK, W. C. 1951. The measurement of vertical transfer of heat and water
vapor in the lower atmosphere. J. Meteorol. 8, 135-145.
SYVERTSEN, J.P., 1984. Light acclimation in citrus leaves. II. CO2 assimilation and light, water, and nitrogen use efficiency. J. Amer. Soc. Hort. Sci 109, 812-817.
SYVERTSEN, J., GRAHAM, J., 1985. Hydraulic conductivity of roots, mineral
nutrition, and leaf gas exchange of citrus rootstocks. J. Am. Soc. Hortic. Sci.
119, 865–869.
SYVERTSEN, J., GOÑI C., OTERO, A., 2003. Fruit load and canopy shading affect
leaf characteristics and net gas exchange of ‘Spring’ navel orange trees.
Tree Physiol. 23, 899–906
SYVERTSEN, J.P., LLOYD, J.J., 1994. Citrus. In: Schaffer, B., Anderson, P.C.
(Eds.) Handbook of environmental physiology of fruit crops. Crc Press, Inc.
USA.
TAN, C.S., BLACK, T.A., NNYAMAH, J.U., 1978. Simple diffusion-model of
transpiration applied to a thinned douglas-fir stand. Ecology 59, 1221–1229
TANNER, B.D., 1988. Use requirements for Bowen ratio and eddy correlation
determination of evapotranspiration. In: Hay, D.R. (Ed.) Planning Now for
Irrigation and Drainage in the 21st Century, Proceedings of Irrigation and
Drainage Division Conference. ASCE, Lincoln, NE, USA, pp. 605–616.
TANNER, B.D., SWIATEK, E., GREEN, J.P., 1993. Density fluctuations and use of
the krypton hygrometer in surface flux measurements. In: Allen, R.G., Neale,
C.M.U. (Eds.), Management of Irrigation and Drainage Systems: Integrated
Perspectives. Proceedings of National Conference on Irrigation and
Drainage Engineering. ASCE, Park City, UT, July, pp. 945–952.
TESTI, L., VILLALOBOS, F.J., ORGAZ, F., 2004. Evapotranspiration of a young
irrigated olive orchard in Southern Spain. Agric. For. Meteorol., 121, 1-18.
© University of Pretoria
141
TESTI, L., ORGAZ, F., VILLALOBOS, F.J., 2006. Variations in bulk canopy
conductance of an irrigated olive (Olea europaea L.) orchard. Environ. Exp.
Bot. 55, 15–28.
TEIXEIRA A. H. DE C., BASTIAANSSEN W. G. M., 2012. Five methods to interpret
field measurements of energy fluxes over a micro-sprinkler-irrigated mango
orchard. Irrig. Sci. 30, 13-28.
TWINE, T.E., KUSTAS, W.P., NORMAN, J.M., COOK, D.R., HOUSER, P.R.,
MEYERS, T.P., PRUEGER, J.H., STARKS, P.J., WESELY, M.L., 2000.
Correcting eddy-covariance flux underestimates over a grassland. Agric. For.
Meteor. 103, 279–300.
VAN BAVEL, C., NEWMAN, J., HILGEMAN, R., 1967. Climate and estimated water
use by an orange orchard. Agric. Meteorol. 4, 27-37
VERTESSY, R.A., BENYON, R.G., O'SUUIVAN, S.K., GRIBBEN, P.R., 1995.
Relationships between stem diameter, sapwood area, leaf area and
transpiration in a young mountain ash forest. Tree Physiol. 15, 559-67.
VILLALOBOS, F.J., ORGAZ F, TESTI L., FERERES E., 2000. Measurement and
modeling of evapotranspiration of olive (Olea europaea L.) orchards. Europ.
J. Agron. 13, 155–163
VILLALOBOS, F.J., TESTI, L., MORENO-PEREZ, M.F., 2009. Evaporation and
canopy conductance of citrus orchards. Agric. Water Manage. 96, 565-573.
VILLALOBOS FJ, TESTI L, ORGAZ F, GARCÍA-TEJERA O, LOPEZ-BERNAL A,
GONZÁLEZ-DUGO MV, BALLESTER-LURBE C, CASTEL JR, ALARCÓN-
CABAÑERO JJ, NICOLÁS-NICOLÁS E, GIRONA J, MARSAL J, FERERES
E, 2013. Modelling canopy conductance and transpiration of fruit trees in
Mediterranean areas: A simplified approach. Agric. For. Meteorol., 171–
172:93-103.
© University of Pretoria
142
VOLSCHENK, T., DE VILLIERS, J.F., BEUKES, O., 2003. The selection and
calibration of a model for irrigation scheduling of deciduous fruit orchards.
Report to the Water Research Commission. Report no. 892/1/03. pp 126.
WARING, R., RUNNING, S., 1978. Sapwood water storage: its contribution to
transpiration and effect upon water conductance through the stems of old-
growth Douglas-fir. Plant, Cell Environ. 1,131-140.
WHITLEY, R., MEDLYN, B., ZEPPEL, M., MACINNIS-NG, C., EAMUS, D., 2009.
Comparing the penman-monteith equation and a modified jarvis-stewart
model with an artificial neural network to estimate stand-scale transpiration
and canopy conductance. J. Hydrol. 373, 256-266.
WIEGAND C.L., SWANSON W.A., 1982. Citrus response to irrigation: I. Irrigation
requirements; daily, monthly and annual evapotranspiration amounts; and
water management recommendations. J. Rio Grande Valley Hort. Soc. 35,
73-85.
WILLIAMS, L.E., AYARS, J.E., 2005. Grapevine water use and the crop coefficient
are linear functions of the shaded area measured beneath the canopy. Agric.
For. Meteorol. 132, 201-211.
WILLIAMS, D., CABLE, W., HULTINE. K., HOEDJES, J., YEPEZ, E.,
SIMONNEAUX, V., ER-RAKI, S., BOULET, G., DE BRUIN, H.,
CHEHBOUNI, A., 2004. Evapotranspiration components determined by
stable isotope, sap flow and eddy covariance techniques. Agric. For.
Meteorol. 125, 241–258
WILSON, K.B., GOLDSTEIN, A.H., FALGE, F., AUBINET, M., BALDOCCHI, D.,
2002. Energy balance closure at FLUXNET sites. Agric. For. Meteorol. 113,
223–243.
WULLSCHLEGER, S.D., MEINZER, F.C., VERTESSY, R.A., 1998. A review of
whole-plant water use studies in trees. Tree Physiol. 18,499.
© University of Pretoria
143
YANG, S., AYDIN, M., YANO, T., LI, X. , 2003. Evapotranspiration of orange trees in
greenhouse lysimeters. Irrig. Sci. 21, 145-149.
YONEMOTO, Y., MATSUMOTO, K., FURUKAWA, T., ASAKAWA, M., OKUDAA, H.,
TAKAHARA, T., 2004. Effects of rootstock and crop load on sap flow rate in
branches of ‘Shirakawa Satsuma’ mandarin (Citrus unshiu Marc.). Sci. Hort.
102, 295–300
ZHANG, H., SIMMONDS, L.P., MORISON, J.I., PAYNE, D., 1997. Estimation of
transpiration by single trees: comparison of sap flow measurements with a
combination equation. Agric. For. Meteorol. 87, 155-169.
ZREIK, C., GONZÁLEZ-ALTOZANO, P., CASTEL, J.R., 2003. Sap flow
determination in plum and mandarin trees by the compensated heat pulse
method. In: Tognetti, R., Raschi, R., (Eds.) Proceedings of the 5Pth
International Workshop on Plant Water Relations and Sap Flux
Measurements. Firenze, Italy, pp 93–108.
© University of Pretoria